SARDet_100K

Name: SARDet_100K
Creator: maas
Published: 2025-12-31 20:24:11
License: 暂无描述

魔搭社区2025-12-31 更新2025-07-12 收录

下载链接：

https://modelscope.cn/datasets/wokaikaixinxin/SARDet_100K

下载链接

链接失效反馈

官方服务：

资源简介：

## Jittor version implementation at [NK-JittorCV/nk-det](https://github.com/NK-JittorCV/nk-det) ## --- ## "SARDet-100K: Towards Open-Source Benchmark and ToolKit for Large-Scale SAR Object Detection" at: [https://arxiv.org/pdf/2403.06534.pdf](https://arxiv.org/pdf/2403.06534.pdf) Yuxuan Li, Xiang Li*, Weijie Li, Qibin Hou, Li Liu, Ming-ming Cheng, Jian Yang* 李宇轩，李翔*，李玮杰，侯淇彬，刘丽，程明明，杨健* [![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/sardet-100k-towards-open-source-benchmark-and/2d-object-detection-on-sardet-100k)](https://paperswithcode.com/sota/2d-object-detection-on-sardet-100k?p=sardet-100k-towards-open-source-benchmark-and) ![MSFA](https://github.com/zcablii/SARDet_100K/blob/main/MSFA/docs/SARDet100K_samples.png) ## Abstract Synthetic Aperture Radar (SAR) object detection has gained significant attention recently due to its irreplaceable all-weather imaging capabilities. However, this research field suffers from both limited public datasets (mostly comprising <2K images with only mono-category objects) and inaccessible source code. To tackle these challenges, we establish a new benchmark dataset and an open-source method for large-scale SAR object detection. Our dataset, SARDet-100K, is a result of intense surveying, collecting, and standardizing 10 existing SAR detection datasets, providing a large-scale and diverse dataset for research purposes. To the best of our knowledge, SARDet-100K is the first COCO-level large-scale multi-class SAR object detection dataset ever created. With this high-quality dataset, we conducted comprehensive experiments and uncovered a crucial challenge in SAR object detection: the substantial disparities between the pretraining on RGB datasets and finetuning on SAR datasets in terms of both data domain and model structure. To bridge these gaps, we propose a novel Multi-Stage with Filter Augmentation (MSFA) pretraining framework that tackles the problems from the perspective of data input, domain transition, and model migration. The proposed MSFA method significantly enhances the performance of SAR object detection models while demonstrating exceptional generalizability and flexibility across diverse models. This work aims to pave the way for further advancements in SAR object detection. ![MSFA](https://github.com/zcablii/SARDet_100K/blob/main/MSFA/docs/SARDet100K.png) ## Introduction This repository is the official site for "SARDet-100K: Towards Open-Source Benchmark and ToolKit for Large-Scale SAR Object Detection" at: [https://arxiv.org/pdf/2403.06534.pdf](https://arxiv.org/pdf/2403.06534.pdf) **DATASET DOWNLOAD at:** (Train, Val, Test) * Baidu Disk: [Dataset](https://pan.baidu.com/s/1dIFOm4V2pM_AjhmkD1-Usw?pwd=SARD) * OneDrive: [Dataset](https://www.kaggle.com/datasets/greatbird/sardet-100k) **Model Weights DOWNLOAD at:** * Baidu Disk: [Models](https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo) * OneDrive: [Models](https://www.kaggle.com/models/greatbird/msfa) **(Only Train and Val sets are released so far.)** <table style="border-collapse: collapse; border: none; border-spacing: 0px;"> <caption> Image and instance level statistics of SARDet-100K dataset. *: Origin datasets are cropped into 512 x 512 patches. </caption> <tr> <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> <b>Dataset</b> <td colspan="4" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> <b>Images</b> <td colspan="4" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> <b>Instances</b> <td rowspan="2" style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Ins/Img <tr> <td style="border-bottom: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> Train <td style="border-bottom: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> Val <td style="border-bottom: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> Test <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> ALL <td style="border-bottom: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> Train <td style="border-bottom: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> Val <td style="border-bottom: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> Test <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> ALL <tr> <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> AIR_SARShip 1*  <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 438 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 23 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 40 <td style="border-right: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 501 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 816 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 33 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 209 <td style="border-right: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 1,058 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 2.11 <tr> <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> AIR_SARShip 2  <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 270 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 15 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 15 <td style="border-right: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 300 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 1,819 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 127 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 94 <td style="border-right: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 2,040 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 6.80 <tr> <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> HRSID  <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 3,642 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 981 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 981 <td style="border-right: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 5,604 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 11,047 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 2,975 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 2,947 <td style="border-right: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 16,969 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 3.03 <tr> <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> MSAR*  <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 27,159 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 1,479 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 1,520 <td style="border-right: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 30,158 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 58,988 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 3,091 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 3,123 <td style="border-right: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 65,202 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 2.16 <tr> <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> SADD  <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 795 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 44 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 44 <td style="border-right: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 883 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 6,891 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 448 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 496 <td style="border-right: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 7,835 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 8.87 <tr> <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> SAR-AIRcraft*  <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 13,976 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 1,923 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 2,989 <td style="border-right: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 18,888 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 27,848 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 4,631 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 5,996 <td style="border-right: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 38,475 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 2.04 <tr> <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> ShipDataset  <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 31,784 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 3,973 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 3,972 <td style="border-right: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 39,729 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 40,761 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 5,080 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 5,044 <td style="border-right: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 50,885 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 1.28 <tr> <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> SSDD  <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 928 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 116 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 116 <td style="border-right: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 1,160 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 2,041 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 252 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 294 <td style="border-right: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 2,587 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 2.23 <tr> <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> OGSOD  <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 14,664 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 1,834 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 1,833 <td style="border-right: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 18,331 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 38,975 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 4,844 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 4,770 <td style="border-right: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 48,589 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 2.65 <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> SIVED  <td style="border-bottom: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 837 <td style="border-bottom: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 104 <td style="border-bottom: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 103 <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 1,044 <td style="border-bottom: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 9,561 <td style="border-bottom: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 1,222 <td style="border-bottom: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 1,230 <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 12,013 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 11.51  <tr> <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> <b>SARDet-100k</b> <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 94,493 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 10,492 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 11,613 <td style="border-right: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 116,598 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 198,747 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 22,703 <td style="text-align: right; padding-right: 3pt; padding-left: 3pt;"> 24,023 <td style="border-right: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 245,653 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 2.11 </table> <table style="border-collapse: collapse; border: none; border-spacing: 0px;"> <caption> SARDet-100K source datasets information. GF-3: Gaofen-3, S-1: Sentinel-1. Target categories S: ship, A: aircraft, C: car, B: bridge, H: harbour, T: tank. </caption> <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> Datasets <td style="border-right: 1px solid black; border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> Target <td style="border-right: 1px solid black; border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> Res. (m) <td style="border-right: 1px solid black; border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> Band <td style="border-right: 1px solid black; border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> Polarization <td style="border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> Satellites <tr> <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> AIR_SARShip <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> S <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 1,3m <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> C <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> VV <td style="padding-right: 3pt; padding-left: 3pt;"> GF-3 <tr> <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> HRSID <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> S <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 0.5~3m <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> C/X <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> HH, HV, VH, VV <td style="padding-right: 3pt; padding-left: 3pt;"> S-1B,TerraSAR-X,TanDEMX <tr> <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> MSAR <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> A, T, B, S <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> < 1m <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> C <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> HH, HV, VH, VV <td style="padding-right: 3pt; padding-left: 3pt;"> HISEA-1 <tr> <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> SADD <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> A <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 0.5~3m <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> X <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> HH <td style="padding-right: 3pt; padding-left: 3pt;"> TerraSAR-X <tr> <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> SAR-AIRcraft <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> A <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 1m <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> C <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> Uni-polar <td style="padding-right: 3pt; padding-left: 3pt;"> GF-3 <tr> <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> ShipDataset <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> S <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 3~25m <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> C <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> HH, VV, VH, HV <td style="padding-right: 3pt; padding-left: 3pt;"> S-1,GF-3 <tr> <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> SSDD <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> S <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 1~15m <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> C/X <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> HH, VV, VH, HV <td style="padding-right: 3pt; padding-left: 3pt;"> S-1,RadarSat-2,TerraSAR-X <tr> <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> OGSOD <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> B, H, T <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 3m <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> C <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> VV/VH <td style="padding-right: 3pt; padding-left: 3pt;"> GF-3 <tr> <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> SIVED <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> C <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 0.1,0.3m <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> Ka,Ku,X <td style="border-right: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> VV/HH <td style="padding-right: 3pt; padding-left: 3pt;"> Airborne SAR synthetic slice </table> ## # MSFA: Multi-Stage with Filter Augmentation pretraining framework ![MSFA](https://github.com/zcablii/SARDet_100K/blob/main/MSFA/docs/MSFA.png) ## Introduction This repository is the official implementation of Multi-Stage with Filter Augmentation (MSFA) pretraining framework in "SARDet-100K: Towards Open-Source Benchmark and ToolKit for Large-Scale SAR Object Detection" Filter Augmentation code is placed under ```MSFA/msfa/models/backbones/MSFA.py```. The code about SARDet-100K dataset is placed under ```MSFA/msfa/datasets/SAR_Det.py```. The train/test configure files used in the main paper are placed under ```local_configs```. ## Results and models ![MSFA_generalizability](https://github.com/zcablii/SARDet_100K/blob/main/MSFA/docs/MSFA_detectors_backbones.png) <table style="border-collapse: collapse; border: none; border-spacing: 0px;"> <caption> Comparison of different pretrain strategies using Faster-RCNN and Res50 as the detection model. </caption> <tr> <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Model Input <td colspan="3" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Pretrain <td rowspan="2" style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> mAP  <td rowspan="2" style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> Config <td rowspan="2" style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> Weight <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Multi-stage <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Dataset <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Component <tr> <td rowspan="4" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> SAR <br> (Raw pixels) <td style="border-right: 1px solid black; border-bottom: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> ✕ <br> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> ImageNet <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Backbone <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 49.0 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\r50_IN_sup\fg_frcnn_IN_sup_sar_r50.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; border-top: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> √<wbr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> ImageNet + DIOR <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Framework <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 49.5 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\r50_dota_pretrain\fg_frcnn_dior_pretrain_sar_r50.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> √<wbr> <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> ImageNet + DOTA <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Backbone <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 49.3 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\r50_dota_backbone_only\fg_frcnn_dota_pretrain_backboneonly_sar_r50.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Framework <td style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> <b>50.2</b> <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\r50_dota_pretrain\fg_frcnn_dota_pretrain_sar_r50.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td rowspan="4" style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> SAR+WST <br> (Filter Augmented) <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> ✕<wbr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> ImageNet <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Backbone <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 49.2 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\r50_IN_sup\fg_frcnn_IN_sup_sar_wavelet_r50.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> √<wbr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> ImageNet + DIOR <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Framework <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 50.1 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\r50_dota_pretrain\fg_frcnn_dior_pretrain_sar_wavelet_r50.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td rowspan="2" style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> √<wbr> <td rowspan="2" style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> ImageNet + DOTA <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Backbone <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 49.6 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\r50_dota_backbone_only\fg_frcnn_dota_pretrain_backboneonly_sar_wavelet_r50.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Framework <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> <b>51.1</b> <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\r50_dota_pretrain\fg_frcnn_dota_pretrain_sar_wavelet_r50.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> </table> ## <table style="border-collapse: collapse; border: none; border-spacing: 0px;"> <caption> Generalization of MSFA on different detection frameworks. IMP: Traditional ImageNet Pretrain on backbone network only. </caption> <tr> <td rowspan="2" colspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Framework <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Pretrain/Model <td colspan="6" style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> Test <td style="border-right: 1px solid rgb(0, 0, 0); border-bottom: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> Config <td style="border-bottom: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> Weight <tr> <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> <b>mAP</b> <td style="border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> @50 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> @75 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> @s <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> @m <td style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> @l <td style="border-right: 1px solid rgb(0, 0, 0); border-top: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <td style="border-top: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <tr> <td rowspan="6" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Two <br> Stage <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Faster RCNN  <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> IMP <td style="padding-right: 3pt; padding-left: 3pt;"> 49.0 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 82.2 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 52.9 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 43.5 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 60.6 <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 55.0 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\r50_IN_sup\fg_frcnn_IN_sup_sar_r50.py"> config </a> <td style="padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> MSFA <td style="border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 51.1<b> (+2.1)</b> <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 83.9 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 54.7 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 45.2 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 62.3 <td style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 57.5 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\r50_dota_pretrain\fg_frcnn_dota_pretrain_sar_wavelet_r50.py"> config </a> <td style="padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Cascade RCNN  <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> IMP <td style="padding-right: 3pt; padding-left: 3pt;"> 51.1 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 81.9 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 55.8 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 44.9 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 62.9 <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 60.3 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_detectors\mrcnn_cascade_IN_sup_r50_sar.py.py"> config </a> <td style="padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> MSFA <td style="border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 53.9 <b> (+2.8)</b> <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 83.4 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 59.8 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 47.2 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 66.1 <td style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 63.2 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_detectors\mrcnn_cascade_r50_dota_pretrained_sar_wavelet.py"> config </a> <td style="padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Grid RCNN  <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> IMP <td style="padding-right: 3pt; padding-left: 3pt;"> 48.8 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 79.1 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 52.9 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 42.4 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 61.9 <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 55.5 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_detectors\grcnn_IN_sup_r50_sar.py"> config </a> <td style="padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> MSFA <td style="border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> <b>51.5 (+2.7)</b> <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 81.7 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 56.3 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 45.1 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 64.1 <td style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 60.0 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_detectors\grcnn_r50_dota_pretrained_sar_wavelet.py"> config </a> <td style="padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td rowspan="7" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Single <br> Stage <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> RetinaNet  <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> IMP <td style="padding-right: 3pt; padding-left: 3pt;"> 47.4 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 79.3 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 49.7 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 40.0 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 59.2 <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 57.5 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_detectors\retinanet_r50_IN_sup_sar.py"> config </a> <td style="padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> MSFA <td style="border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 49.0<b> (+1.6)</b> <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 80.1 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 52.6 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 41.3 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 61.1 <td style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 59.4 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_detectors\retinanet_r50_dota_pretrained_sar_wavelet.py"> config </a> <td style="padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td rowspan="3" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> GFL  <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> IMP <td style="padding-right: 3pt; padding-left: 3pt;"> 49.8 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 80.9 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 53.3 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 42.3 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 62.4 <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 58.1 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_detectors\gfl_r50_IN_sup_sar.py"> config </a> <td style="padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> MSFA <td style="border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 53.7<b> (+3.9)</b> <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 84.2 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 57.8 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 47.8 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 66.2 <td style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 59.5 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_detectors\gfl_r50_dota_pretrained_sar_wavelet.py"> config </a> <td style="padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> DenoDet <td style="border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 55.4<b> (+5.6)</b> <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 84.7 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 58.3 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 49.5 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 67.6 <td style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 63.2 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://github.com/zcablii/SARDet_100K/blob/main/MSFA/local_configs/DeonDet/gfl_r50_denodet_sardet.py"> config </a> <td style="padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> FCOS  <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> IMP <td style="padding-right: 3pt; padding-left: 3pt;"> 46.5 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 80.9 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 49.0 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 41.1 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 59.2 <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 50.4 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_detectors\fcos_r50_IN_sup_sar.py"> config </a> <td style="padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> MSFA <td style="border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 48.5<b> (+2.0)</b> <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 82.1 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 51.4 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 42.9 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 60.4 <td style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 56.0 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_detectors\fcos_r50_dota_pretrained_sar_wavelet.py"> config </a> <td style="padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td rowspan="6" style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> End to <br> End <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> DETR  <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> IMP <td style="padding-right: 3pt; padding-left: 3pt;"> 31.8 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 62.3 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 30.0 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 22.2 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 44.9 <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 41.1 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_detectors\detr_r50_IN_sup_sar.py"> config </a> <td style="padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> MSFA <td style="border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 47.2<b> (+15.4)</b> <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 77.5 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 49.8 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 37.9 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 62.9 <td style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 58.2 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_detectors\detr_r50_dota_pretrained_sar_wavelet.py"> config </a> <td style="padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Deformable DETR  <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> IMP <td style="padding-right: 3pt; padding-left: 3pt;"> 50.0 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 85.1 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 51.7 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 44.0 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 65.1 <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 61.2 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_detectors\deformable-detr_r50_IN_sup_sar.py"> config </a> <td style="padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> MSFA <td style="border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 51.3<b> (+1.3)</b> <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 85.3 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 54.0 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 44.9 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 65.6 <td style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 61.7 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_detectors\deformable-detr_r50_dota_pretrain_sar_wavelet.py"> config </a> <td style="padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Sparse RCNN  <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> IMP <td style="padding-right: 3pt; padding-left: 3pt;"> 38.1 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 68.8 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 38.8 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 29.0 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 51.3 <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 48.7 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_detectors\sparse-rcnn_r50_IN_spu_sar.py"> config </a> <td style="padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> MSFA <td style="border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 41.4<b> (+3.3)</b> <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 74.1 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 41.8 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 33.6 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 53.9 <td style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 53.4 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_detectors\sparse-rcnn_r50_dota_pretrain_sar_wavelet.py"> config </a> <td style="padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> <td rowspan="2" style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Dab-DETR  <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> IMP <td style="padding-right: 3pt; padding-left: 3pt;"> 45.9 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 79.0 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 47.9 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 38.0 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 61.1 <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 55.0 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_detectors\dab-detr_r50_IN_sup_sar.py"> config </a> <td style="padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> MSFA <td style="padding-right: 3pt; padding-left: 3pt;"> 48.2<b> (+2.3)</b> <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 81.1 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 51.0 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 41.2 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 63.1 <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 55.4 <td style="border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_detectors\dab-detr_r50_dota_pretrain_sar_wavelet.py"> config </a> <td style="padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> </table> ## <table style="border-collapse: collapse; border: none; border-spacing: 0px;"> <caption> Generalization of MSFA on different detection backbones. IMP: Traditional ImageNet Pretrain on backbone network only. </caption> <tr> <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Framework <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> #P(M) <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Pretrain <td colspan="6" style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> Test <td style="border-left: 1px solid rgb(0, 0, 0); border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> Config <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> Weight <tr> <td style="border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> <b>mAP</b> <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> @50 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> @75 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> @s <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> @m <td style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> @l <td style="border-left: 1px solid rgb(0, 0, 0); border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <tr> <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> R50  <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 25.6 <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> IMP <td style="padding-right: 3pt; padding-left: 3pt;"> 49.0 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 82.2 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 52.9 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 43.5 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 60.6 <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 55.0 <td style="border-left: 1px solid rgb(0, 0, 0); border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\r50_IN_sup\fg_frcnn_IN_sup_sar_r50.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> MSFA <td style="border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 51.1<b> (+2.1)</b> <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 83.9 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 54.7 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 45.2 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 62.3 <td style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 57.5 <td style="border-left: 1px solid rgb(0, 0, 0); border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\r50_dota_pretrain\fg_frcnn_dota_pretrain_sar_wavelet_r50.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> R101  <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 44.7 <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> IMP <td style="padding-right: 3pt; padding-left: 3pt;"> 51.2 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 84.1 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 55.6 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 45.9 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 61.9 <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 56.3 <td style="border-left: 1px solid rgb(0, 0, 0); border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_backbones\fg_frcnn_IN_sup_sar_r101.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> MSFA <td style="border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 52.0<b> (+0.8)</b> <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 84.6 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 56.6 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 46.6 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 63.4 <td style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 57.7 <td style="border-left: 1px solid rgb(0, 0, 0); border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_backbones\fg_frcnn_dota_pretrain_sar_r101_wavelet.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> R152  <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 60.2 <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> IMP <td style="padding-right: 3pt; padding-left: 3pt;"> 51.9 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 85.2 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 55.9 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 46.4 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 62.5 <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 57.9 <td style="border-left: 1px solid rgb(0, 0, 0); border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_backbones\fg_frcnn_IN_sup_sar_r152.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> MSFA <td style="border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 52.4<b> (+0.5)</b> <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 85.4 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 57.2 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 47.4 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 63.3 <td style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 58.7 <td style="border-left: 1px solid rgb(0, 0, 0); border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_backbones\fg_frcnn_dota_pretrain_sar_r152_wavelet.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> ConvNext-T  <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 28.6 <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> IMP <td style="padding-right: 3pt; padding-left: 3pt;"> 53.2 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 86.3 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 58.1 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 47.2 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 65.2 <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 59.6 <td style="border-left: 1px solid rgb(0, 0, 0); border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_backbones\fg_frcnn_IN_sup_sar_convnext_t.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> MSFA <td style="border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 54.8<b> (+1.6)</b> <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 87.1 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 59.8 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 48.8 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 66.7 <td style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 62.1 <td style="border-left: 1px solid rgb(0, 0, 0); border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_backbones\fg_frcnn_dota_pretrain_sar_convnext_t_wavelet.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> ConvNext-S  <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 50.1 <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> IMP <td style="padding-right: 3pt; padding-left: 3pt;"> 54.2 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 87.8 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 59.2 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 49.2 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 65.8 <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 59.8 <td style="border-left: 1px solid rgb(0, 0, 0); border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_backbones\fg_frcnn_IN_sup_sar_convnext_s.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> MSFA <td style="border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 55.4<b> (+1.2)</b> <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 87.6 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 60.7 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 50.1 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 67.1 <td style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 61.3 <td style="border-left: 1px solid rgb(0, 0, 0); border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_backbones\fg_frcnn_dota_pretrain_sar_convnext_s_wavelet.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> ConvNext-B  <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 88.6 <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> IMP <td style="padding-right: 3pt; padding-left: 3pt;"> 55.1 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 87.8 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 59.5 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 48.9 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 66.9 <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 61.1 <td style="border-left: 1px solid rgb(0, 0, 0); border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_backbones\fg_frcnn_IN_sup_sar_convnext_b.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> MSFA <td style="border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 56.4<b> (+1.3)</b> <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 88.2 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 61.5 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 51.1 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 68.3 <td style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 62.4 <td style="border-left: 1px solid rgb(0, 0, 0); border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_backbones\fg_frcnn_dota_pretrain_sar_convnext_b_wavelet.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> VAN-T  <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;">  4.1 <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> IMP <td style="padding-right: 3pt; padding-left: 3pt;"> 45.8 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 79.8 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 48.0 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 38.6 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 57.9 <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 53.3 <td style="border-left: 1px solid rgb(0, 0, 0); border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_backbones\fg_frcnn_IN_sup_sar_van_t.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> MSFA <td style="border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 47.6 <b>(+1.8)</b> <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 81.4 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 50.6 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 40.5 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 59.4 <td style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 56.7 <td style="border-left: 1px solid rgb(0, 0, 0); border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_backbones\fg_frcnn_dota_pretrain_sar_van_t_wavelet.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> VAN-S  <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 13.9 <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> IMP <td style="padding-right: 3pt; padding-left: 3pt;"> 49.5 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 83.8 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 52.8 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 43.2 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 61.6 <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 56.4 <td style="border-left: 1px solid rgb(0, 0, 0); border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_backbones\fg_frcnn_IN_sup_sar_van_s.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> MSFA <td style="border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 51.5<b> (+2.0)</b> <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 85.0 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 55.6 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 44.8 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 63.4 <td style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 60.4 <td style="border-left: 1px solid rgb(0, 0, 0); border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_backbones\fg_frcnn_dota_pretrain_sar_van_s_wavelet.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> VAN-B  <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 26.6 <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> IMP <td style="padding-right: 3pt; padding-left: 3pt;"> 53.5 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 86.8 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 58.0 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 47.3 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 65.5 <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 60.6 <td style="border-left: 1px solid rgb(0, 0, 0); border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_backbones\fg_frcnn_IN_sup_sar_van_b.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> MSFA <td style="border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 55.1<b> (+1.6)</b> <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 87.7 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 60.2 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 48.8 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 67.3 <td style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 62.2 <td style="border-left: 1px solid rgb(0, 0, 0); border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_backbones\fg_frcnn_dota_pretrain_sar_van_b_wavelet.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Swin-T  <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 28.3 <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> IMP <td style="padding-right: 3pt; padding-left: 3pt;"> 48.4 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 83.5 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 50.8 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 42.8 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 59.7 <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 55.7 <td style="border-left: 1px solid rgb(0, 0, 0); border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_backbones\fg_frcnn_IN_sup_sar_swin_t.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> MSFA <td style="border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 50.2<b> (+1.8)</b> <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 84.1 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 53.9 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 44.1 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 61.3 <td style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 58.8 <td style="border-left: 1px solid rgb(0, 0, 0); border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_backbones\fg_frcnn_dota_pretrain_sar_swin_t_wavelet.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Swin-S  <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 49.6 <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> IMP <td style="padding-right: 3pt; padding-left: 3pt;"> 53.1 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 87.3 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 57.8 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 47.4 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 63.9 <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 60.6 <td style="border-left: 1px solid rgb(0, 0, 0); border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_backbones\fg_frcnn_IN_sup_sar_swin_s.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> MSFA <td style="border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 54.0<b> (+0.9) </b> <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 87.0 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 59.2 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 48.2 <td style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 64.5 <td style="border-bottom: 1px solid black; border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 61.9 <td style="border-left: 1px solid rgb(0, 0, 0); border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_backbones\fg_frcnn_dota_pretrain_sar_swin_s_wavelet.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td rowspan="2" style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> Swin-B  <td rowspan="2" style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 87.8 <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> IMP <td style="padding-right: 3pt; padding-left: 3pt;"> 53.8 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 87.8 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 59.0 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 49.1 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 64.6 <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 60.0 <td style="border-left: 1px solid rgb(0, 0, 0); border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_backbones\fg_frcnn_IN_sup_sar_swin_b.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> <tr> <td style="border-right: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> MSFA <td style="padding-right: 3pt; padding-left: 3pt;"> 55.7<b> (+1.9)</b> <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 87.8 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 61.4 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 50.5 <td style="text-align: center; padding-right: 3pt; padding-left: 3pt;"> 66.5 <td style="border-right: 1px solid rgb(0, 0, 0); text-align: center; padding-right: 3pt; padding-left: 3pt;"> 62.5 <td style="border-left: 1px solid rgb(0, 0, 0); border-right: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="local_configs\SARDet\other_backbones\fg_frcnn_dota_pretrain_sar_swin_b_wavelet.py"> config </a> <td style="border-left: 1px solid rgb(0, 0, 0); padding-right: 3pt; padding-left: 3pt;"> <a href="https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo"> weight </a> </table> ## Installation Our code depends on [PyTorch](https://pytorch.org/), [MMCV](https://github.com/open-mmlab/mmcv) and [MMDetection](https://github.com/open-mmlab/mmdetection). Below are quick steps for installation. Please refer to [Install Guide](https://mmrotate.readthedocs.io/en/latest/install.html) for more detailed instruction. ``` # change directory into the project main code cd MSFA # create env conda create -y -n MSFA python=3.8 conda activate MSFA # install pytorch conda install pytorch==2.0.1 torchvision==0.15.2 torchaudio==2.0.2 pytorch-cuda=11.8 -c pytorch -c nvidia # or pip install torch==2.0.1 torchvision==0.15.2 torchaudio==2.0.2 --index-url https://download.pytorch.org/whl/cu118 # install dependencies of openmmlab pip install -U openmim mim install "mmengine==0.8.4" mim install "mmcv==2.0.1" mim install "mmdet==3.1.0" # install other dependencies pip install -r requirements.txt # install MSFA pip install -v -e . ``` ## Get Started Please see [get_started.md](https://github.com/open-mmlab/mmdetection/blob/main/docs/en/get_started.md) for the basic usage of MMDetection. ## Acknowledgement We extend our deepest gratitude to Bo Zhang, Chenglong Li, Tian Tian, Tianwen Zhang, Xiaoling Zhang (ordered alphabetically by first name) and numerous other researchers for permitting us to integrate their datasets. Their contributions have significantly advanced and promoted research in this field ## Citation If you use this toolbox or benchmark in your research, please cite this project. ```bibtex @inproceedings{li2024sardet100k, title={SARDet-100K: Towards Open-Source Benchmark and ToolKit for Large-Scale SAR Object Detection}, author={Yuxuan Li and Xiang Li and Weijie Li and Qibin Hou and Li Liu and Ming-Ming Cheng and Jian Yang}, year={2024}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems (NeurIPS)}, } @article{zhang2025rsar, title={RSAR: Restricted State Angle Resolver and Rotated SAR Benchmark}, author={Zhang, Xin and Yang, Xue and Li, Yuxuan and Yang, Jian and Cheng, Ming-Ming and Li, Xiang}, journal={Proceedings of the IEEE/CVF conference on computer vision and pattern recognition}, year={2025} } @article{dai2024denodet, title={DenoDet: Attention as Deformable Multi-Subspace Feature Denoising for Target Detection in SAR Images}, author={Dai, Yimian and Zou, Minrui and Li, Yuxuan and Li, Xiang and Ni, Kang and Yang, Jian}, journal={arXiv preprint arXiv:2406.02833}, year={2024} } ``` ## Star History [![Star History Chart](https://api.star-history.com/svg?repos=zcablii/SARDet_100K&type=Date)](https://star-history.com/#zcablii/SARDet_100K&Date) ## License This project is released under the [Attribution-NonCommercial 4.0 International](LICENSE).

## 基于计图（Jittor）的实现仓库：[NK-JittorCV/nk-det](https://github.com/NK-JittorCV/nk-det) ## --- ## 《SARDet-100K：面向大规模合成孔径雷达目标检测的开源基准与工具集》，论文链接：[https://arxiv.org/pdf/2403.06534.pdf](https://arxiv.org/pdf/2403.06534.pdf) 作者：李宇轩，李翔*，李玮杰，侯淇彬，刘丽，程明明，杨健* [![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/sardet-100k-towards-open-source-benchmark-and/2d-object-detection-on-sardet-100k)](https://paperswithcode.com/sota/2d-object-detection-on-sardet-100k?p=sardet-100k-towards-open-source-benchmark-and) ![MSFA](https://github.com/zcablii/SARDet_100K/blob/main/MSFA/docs/SARDet100K_samples.png) ## 摘要合成孔径雷达（Synthetic Aperture Radar, SAR）目标检测凭借其不可替代的全天候成像能力，近年来受到了广泛关注。然而该研究领域面临两大困境：公开数据集数量有限（绝大多数仅包含不足2000张图像，且仅含单类别目标），且源代码难以获取。为解决这些挑战，我们构建了面向大规模SAR目标检测的全新基准数据集与开源方法。我们的数据集SARDet-100K通过对10个现有SAR检测数据集进行系统性调研、收集与标准化处理而成，可为研究提供大规模且多样化的数据支持。据我们所知，SARDet-100K是首个达到COCO级别的大规模多类别SAR目标检测数据集。借助该高质量数据集，我们开展了全面的实验，并揭示了SAR目标检测领域的一项关键挑战：在数据域与模型结构两个层面，基于RGB数据集的预训练与基于SAR数据集的微调之间存在显著差异。为弥补这些差距，我们提出了一种新颖的**滤波增强多阶段（Multi-Stage with Filter Augmentation, MSFA）**预训练框架，从数据输入、域迁移与模型适配三个维度解决上述问题。所提MSFA方法显著提升了SAR目标检测模型的性能，同时在多种不同模型上展现出优异的泛化性与灵活性。本研究旨在为SAR目标检测领域的进一步发展奠定基础。 ![MSFA](https://github.com/zcablii/SARDet_100K/blob/main/MSFA/docs/SARDet100K.png) ## 简介本仓库为论文《SARDet-100K：面向大规模SAR目标检测的开源基准与工具集》的官方实现站点，论文链接：[https://arxiv.org/pdf/2403.06534.pdf](https://arxiv.org/pdf/2403.06534.pdf) **数据集下载（训练集、验证集、测试集）：** * 百度网盘：[数据集](https://pan.baidu.com/s/1dIFOm4V2pM_AjhmkD1-Usw?pwd=SARD) * OneDrive：[数据集](https://www.kaggle.com/datasets/greatbird/sardet-100k) **模型权重下载：** * 百度网盘：[模型权重](https://pan.baidu.com/s/1SuEOl_ImqjoT5Y3pYxZt4w?pwd=c6fo) * OneDrive：[模型权重](https://www.kaggle.com/models/greatbird/msfa) **（截至目前仅开放训练集与验证集下载。）** <table style="border-collapse: collapse; border: none; border-spacing: 0px;"> <caption> SARDet-100K数据集的图像与实例级统计信息。*：原始数据集被裁剪为512×512的图像块。 </caption> <tr> <td rowspan="2" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> <b>数据集</b> <td colspan="4" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> <b>图像数量</b> <td colspan="4" style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> <b>实例数量</b> <td rowspan="2" style="border-bottom: 1px solid black; text-align: center; padding-right: 3pt; padding-left: 3pt;"> 每张图像的实例数 <tr> <td style="border-bottom: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 训练集 <td style="border-bottom: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 验证集 <td style="border-bottom: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 测试集 <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 总计 <td style="border-bottom: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 训练集 <td style="border-bottom: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 验证集 <td style="border-bottom: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 测试集 <td style="border-right: 1px solid black; border-bottom: 1px solid black; text-align: right; padding-right: 3pt; padding-left: 3pt;"> 总计 </table> <table style="border-collapse: collapse; border: none; border-spacing: 0px;"> <caption> SARDet-100K的源数据集信息。GF-3：高分三号卫星，S-1：哨兵一号卫星。目标类别：S（船舶）、A（飞机）、C（车辆）、B（桥梁）、H（港口）、T（坦克）。 </caption> <tr> <td style="border-right: 1px solid black; border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 数据集 <td style="border-right: 1px solid black; border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 目标类别 <td style="border-right: 1px solid black; border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 空间分辨率(m) <td style="border-right: 1px solid black; border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 波段 <td style="border-right: 1px solid black; border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 极化方式 <td style="border-bottom: 1px solid black; padding-right: 3pt; padding-left: 3pt;"> 搭载卫星 </table> ## MSFA：滤波增强多阶段预训练框架 ![MSFA](https://github.com/zcablii/SARDet_100K/blob/main/MSFA/docs/MSFA.png) ## 简介本仓库为论文《SARDet-100K：面向大规模SAR目标检测的开源基准与工具集》中提出的滤波增强多阶段预训练框架（MSFA）的官方实现。滤波增强模块的代码位于`MSFA/msfa/models/backbones/MSFA.py`。与SARDet-100K数据集相关的代码位于`MSFA/msfa/datasets/SAR_Det.py`。主论文中使用的训练/测试配置文件位于`local_configs`目录下。 ## 实验结果与模型 ![MSFA_generalizability](https://github.com/zcablii/SARDet_100K/blob/main/MSFA/docs/MSFA_detectors_backbones.png) <table style="border-collapse: collapse; border: none; border-spacing: 0px;"> <caption> 以Faster RCNN与ResNet50作为检测模型的不同预训练策略对比。 </caption> </table> <table style="border-collapse: collapse; border: none; border-spacing: 0px;"> <caption> MSFA在不同检测框架上的泛化性。IMP：仅基于ImageNet的主干网络传统预训练。 </caption> </table> <table style="border-collapse: collapse; border: none; border-spacing: 0px;"> <caption> MSFA在不同检测主干网络上的泛化性。IMP：仅基于ImageNet的主干网络传统预训练。 </caption> </table> ## 安装本代码依赖[PyTorch](https://pytorch.org/)、[MMCV](https://github.com/open-mmlab/mmcv)与[MMDetection](https://github.com/open-mmlab/mmdetection)。以下为快速安装步骤，详细安装指南请参考[安装文档](https://mmrotate.readthedocs.io/en/latest/install.html)。 # 切换至项目主目录 cd MSFA # 创建并激活虚拟环境 conda create -y -n MSFA python=3.8 conda activate MSFA # 安装PyTorch conda install pytorch==2.0.1 torchvision==0.15.2 torchaudio==2.0.2 pytorch-cuda=11.8 -c pytorch -c nvidia # 或使用pip安装 pip install torch==2.0.1 torchvision==0.15.2 torchaudio==2.0.2 --index-url https://download.pytorch.org/whl/cu118 # 安装OpenMMLab相关依赖 pip install -U openmim mim install "mmengine==0.8.4" mim install "mmcv==2.0.1" mim install "mmdet==3.1.0" # 安装其他依赖项 pip install -r requirements.txt # 本地安装MSFA pip install -v -e . ## 快速入门关于MMDetection的基础使用方法，请参考[get_started.md](https://github.com/open-mmlab/mmdetection/blob/main/docs/en/get_started.md)。 ## 致谢我们衷心感谢Bo Zhang、Chenglong Li、Tian Tian、Tianwen Zhang、Xiaoling Zhang（按姓氏首字母排序）以及众多其他研究者允许我们整合他们的数据集。他们的贡献极大地推动了本领域的研究发展。 ## 引用如果您在研究中使用了本工具包或基准数据集，请引用该项目。 bibtex @inproceedings{li2024sardet100k, title={SARDet-100K: Towards Open-Source Benchmark and ToolKit for Large-Scale SAR Object Detection}, author={Yuxuan Li and Xiang Li and Weijie Li and Qibin Hou and Li Liu and Ming-Ming Cheng and Jian Yang}, year={2024}, booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems (NeurIPS)}, } @article{zhang2025rsar, title={RSAR: Restricted State Angle Resolver and Rotated SAR Benchmark}, author={Zhang, Xin and Yang, Xue and Li, Yuxuan and Yang, Jian and Cheng, Ming-Ming and Li, Xiang}, journal={Proceedings of the IEEE/CVF conference on computer vision and pattern recognition}, year={2025} } @article{dai2024denodet, title={DenoDet: Attention as Deformable Multi-Subspace Feature Denoising for Target Detection in SAR Images}, author={Dai, Yimian and Zou, Minrui and Li, Yuxuan and Li, Xiang and Ni, Kang and Yang, Jian}, journal={arXiv preprint arXiv:2406.02833}, year={2024} } ## 星标历史 [![Star History Chart](https://api.star-history.com/svg?repos=zcablii/SARDet_100K&type=Date)](https://star-history.com/#zcablii/SARDet_100K&Date) ## 许可证本项目采用[署名-非商业性使用4.0国际（Attribution-NonCommercial 4.0 International）](LICENSE)许可证发布。

提供机构：

maas

创建时间：

2025-07-10

搜集汇总

数据集介绍

以上内容由遇见数据集搜集并总结生成

5,000+

优质数据集

54 个

任务类型

进入经典数据集