DETR for Crowd Pedestrian Detection
This work addresses the problem of improving end-to-end detectors for crowd pedestrian detection, which is a challenging task due to high occlusion and overlap, for researchers and practitioners working on computer vision and autonomous systems.
This paper investigates the performance of end-to-end detectors (EDs) like DETR and deformable DETR for crowd pedestrian detection, finding them surprisingly underperforming compared to Faster-RCNN. The authors identify the reasons for this poor performance and propose a new decoder, a mechanism to leverage visible pedestrian parts, and a faster bipartite matching algorithm, resulting in a new detector (PED) that outperforms previous EDs and Faster-RCNN on CityPersons and CrowdHuman.
Pedestrian detection in crowd scenes poses a challenging problem due to the heuristic defined mapping from anchors to pedestrians and the conflict between NMS and highly overlapped pedestrians. The recently proposed end-to-end detectors(ED), DETR and deformable DETR, replace hand designed components such as NMS and anchors using the transformer architecture, which gets rid of duplicate predictions by computing all pairwise interactions between queries. Inspired by these works, we explore their performance on crowd pedestrian detection. Surprisingly, compared to Faster-RCNN with FPN, the results are opposite to those obtained on COCO. Furthermore, the bipartite match of ED harms the training efficiency due to the large ground truth number in crowd scenes. In this work, we identify the underlying motives driving ED's poor performance and propose a new decoder to address them. Moreover, we design a mechanism to leverage the less occluded visible parts of pedestrian specifically for ED, and achieve further improvements. A faster bipartite match algorithm is also introduced to make ED training on crowd dataset more practical. The proposed detector PED(Pedestrian End-to-end Detector) outperforms both previous EDs and the baseline Faster-RCNN on CityPersons and CrowdHuman. It also achieves comparable performance with state-of-the-art pedestrian detection methods. Code will be released soon.