Visible Feature Guidance for Crowd Pedestrian Detection

TL;DR

This paper introduces Visible Feature Guidance (VFG), a novel approach for crowd pedestrian detection that improves bounding box accuracy and association in occluded, dense scenes, enhancing detection performance across multiple datasets.

Contribution

The paper proposes VFG, a new mechanism that uses visible features for bounding box regression and NMS, and employs Hungarian algorithm for part association, improving detection accuracy in crowded scenes.

Findings

Achieves 2-3% improvements in mAP and AP50 on multiple datasets.

More effective in stricter IoU conditions, especially MR-2.

Provides a strong benchmark for parts association in crowd detection.

Abstract

Heavy occlusion and dense gathering in crowd scene make pedestrian detection become a challenging problem, because it's difficult to guess a precise full bounding box according to the invisible human part. To crack this nut, we propose a mechanism called Visible Feature Guidance (VFG) for both training and inference. During training, we adopt visible feature to regress the simultaneous outputs of visible bounding box and full bounding box. Then we perform NMS only on visible bounding boxes to achieve the best fitting full box in inference. This manner can alleviate the incapable influence brought by NMS in crowd scene and make full bounding box more precisely. Furthermore, in order to ease feature association in the post application process, such as pedestrian tracking, we apply Hungarian algorithm to associate parts for a human instance. Our proposed method can stably bring about 2~3% improvements in mAP and AP50 for both two-stage and one-stage detector. It's also more effective for MR-2 especially with the stricter IoU. Experiments on Crowdhuman, Cityperson, Caltech and KITTI datasets show that visible feature guidance can help detector achieve promisingly better performances. Moreover, parts association produces a strong benchmark on Crowdhuman for the vision community.