R2C7K

该数据集包含6,615张图片，涵盖64个类别，旨在促进对伪装目标检测（RefCOD）的研究。其中分为Camo子集和Ref子集，同时提供了参照分割，包括训练和测试图片。该数据集规模较大，针对的任务是参照伪装目标检测。This dataset contains 6,615 images spanning 64 categories, aiming to advance research on referential camouflaged object detection (RefCOD). It is divided into the Camo subset and the Ref subset, with referential segmentation annotations provided for both training and test images. With a considerable scale, this dataset is specifically designed for the task of referential camouflaged object detection.

搜集汇总

背景与挑战

背景概述

R2C7K数据集是一个专门用于指向性伪装物体检测的数据集，包含64个类别的伪装物体，分为训练集和测试集。数据集提供了图像、对象表示和前景图，支持R2CNet框架的开发和评估。