Adaptive NMS: Refining Pedestrian Detection in a Crowd

2019/04/07 by Songtao Liu, Di Huang, Liu, Songtao +3 · 5 citations
Computer Science · #Video Surveillance and Tracking Methods #Anomaly Detection Techniques and Applications #Speech and Audio Processing

paper · pdf · doi:10.48550/arxiv.1904.03629

Abstract

Pedestrian detection in a crowd is a very challenging issue. This paper addresses this problem by a novel Non-Maximum Suppression (NMS) algorithm to better refine the bounding boxes given by detectors. The contributions are threefold: (1) we propose adaptive-NMS, which applies a dynamic suppression threshold to an instance, according to the target density; (2) we design an efficient subnetwork to learn density scores, which can be conveniently embedded into both the single-stage and two-stage detectors; and (3) we achieve state of the art results on the CityPersons and CrowdHuman benchmarks.

Citations

• Local Decorrelation For Improved Detection
• Very Deep Convolutional Networks for Large-Scale Image Recognition
• CrowdHuman: A Benchmark for Detecting Human in a Crowd
• Robust Real-Time Face Detection

Cited by

• Group Evidence Matters: Tiling-based Semantic Gating for Dense Object Detection
• CrowdQuery: Density-Guided Query Module for Enhanced 2D and 3D Detection in Crowded Scenes
• CMF-IoU: Multi-Stage Cross-Modal Fusion 3D Object Detection with IoU Joint Prediction
• PSC-Net: Learning Part Spatial Co-occurrence for Occluded Pedestrian Detection
• A Late Collaborative Perception Framework for 3D Multi-Object and Multi-Source Association and Fusion

Related

• NMS by Representative Region: Towards Crowded Pedestrian Detection by Proposal Pairing
• A convnet for non-maximum suppression
• DETR for Crowd Pedestrian Detection
• Visible Feature Guidance for Crowd Pedestrian Detection
• Resisting Crowd Occlusion and Hard Negatives for Pedestrian Detection in the Wild