CompetitionRelative

1. 目标检测与追踪

• 10亿像素的图像视频中进行目标检测与跟踪

• 任务一– 目标检测任务：在PANDA数据集上完成车辆检测和行人检测，其中行人检测需要检测出可见范围框、全身范围框和头部范围框，车辆检测只需要检测出可见范围框。

• 任务二– 多目标跟踪任务：输出PANDA视频序列上的行人轨迹，即输出每帧图片的行人ID和位置框。

• https://github.com/hasanirtiza/Pedestron，对应论文《Generalizable Pedestrian Detection: The Elephant In The Room》，基于MMDetection框架的行人检测代码仓库。
• https://github.com/saic-vul/iterdet，拥挤环境的目标检测，对应论文《IterDet: Iterative Scheme for Object Detection in Crowded Environments》，同样是基MMDetection框架。
• https://github.com/liuwei16/CSP，对应论文《High-level Semantic Feature Detection: A New Perspective for Pedestrian Detection》。
• PS-RCNN: Detecting Secondary Human Instances in a Crowd via Primary Object Suppression
• NMS by Representative Region: Towards Crowded Pedestrian Detection by Proposal Pairing
• Double Anchor R-CNN for Human Detection in a Crowd