Awesome-Pedestrian
收藏官方服务：
资源简介：
这是一个关于行人相关资源的合集，主要收集和整理行人检测、属性识别等领域的数据集、论文、开源项目、挑战赛和在线资源。合集覆盖了多个行人数据集，如CityPersons、Caltech、CrowdHuman等，并提供了SOTA方法、评估指标和相关文档，旨在为行人研究提供全面的参考。
This is a curated collection of pedestrian-related resources, focusing on collecting and organizing datasets, papers, open-source projects, challenges and online resources in the fields of pedestrian detection and attribute recognition. The collection covers multiple pedestrian datasets such as CityPersons, Caltech, CrowdHuman and others, and provides SOTA methods, evaluation metrics and relevant documents, aiming to offer a comprehensive reference for pedestrian research.
创建时间：
2019-09-04
原始信息汇总
数据集与SOTA
行人检测数据集：
- CityPersons (2.975k)
- Caltech (42.782k)
- CrowdHuman (15k)
- KITTI (3.712k)
- COCOPersons (64.115k)
- OCHuman (4.731k)
- WiderPerson (13.382k)
- INRIA
- NICTA
- KITTI
- CUHK Occlusion Dataset
- CUHK Square Dataset
- BIWI Walking Pedestrians dataset
- Central Pedestrian Crossing Sequences
- TUD
- KAIST multispectral dataset
SOTA参考榜单：
- Caltech 排行榜
- CityPersons 排行榜
挑战赛/竞赛
- WIDER Face & Person Challenge
- MOTChallenge: Detection in Crowded Scenes
- CrowdHuman
评估指标
- mAP
- MR (Reasonable)
- MR (Reasonable_small)
- MR (Reasonable_occ=heavy)
- MR (All)
开源项目
- MMDetection
- maskrcnn-benchmark
- CSP (PyTorch & Keras)
- Bi-box_Regression
- ALFNet
- Repulsion_Loss
- SDS-RCNN
在线资源
- Pedestrian-Detection
- awesome-pedestrian-detection
- Pedestrian-Attribute-Recognition-Paper-List
- human_papernotes
论文与文档
人群/遮挡相关论文（按会议分类）：
- CVPR 2019：Autoregressive Network Phases、High-Level Semantic Feature Detection、Adaptive NMS
- ECCV 2018：Occlusion-aware R-CNN、Bi-box Regression、Graininess-Aware Deep Feature Learning、Small-scale Pedestrian Detection、Asymptotic Localization Fitting
- CVPR 2018：Improving Occlusion、Guided Attention、Repulsion Loss、WILDTRACK
- ICCV 2017：Multi-label Learning of Part Detectors、Simultaneous Detection & Segmentation
- CVPR 2017：CityPersons、Cross-Modal Deep Representations、What Can Help Pedestrian Detection?、Progressive Latent Model、Adversarial Imposters
- ECCV 2016：Is Faster R-CNN Doing Well for Pedestrian Detection?
- CVPR 2016：Semantic Channels、How Far are We from Solving Pedestrian Detection?、Appearance Constancy and Shape Symmetry
- Arxiv 2019及IEEE Access历年论文
轨迹预测相关论文：
- CVPR 2019：SR-LSTM
- CVPR 2018：Encoding Crowd Interaction
- CVPR 2017：Forecasting Interactive Dynamics
人群计数相关论文：
- CVPR 2019：Point in, Box out
- Arxiv 2019：Dynamic Region Division
属性/分析相关论文：
- Arxiv 2019：Detector-in-Detector、Attribute Aware Pooling、Pedestrian Attribute Recognition: A Survey
- CVPR 2017：HydraPlus-Net
