Abstract:With the development of self-driving technology, pedestrian detection models are becoming more useful and important. Although object detection models based on Convolutional Neural Networks have achieved favorable detection results, due to the occlusion in crowded scenarios, dense pedestrian detection is still a challenging task compared with generic object detection. For a CNN-based object detector, an appropriate label assignment method is significant. In this paper, we propose an effective training sample allocation method called Occlusion-Prediction aware Label Assignment (OPLA) to improve the performance of the detector in crowded scenarios. For those Ground Truths (GTs) without severe occlusion, we directly assign prediction boxes with the best matching quality as the corresponding positive samples. In terms of GTs in heavily crowded areas, we fine-tune the matching quality based on visible ratio to enhance the training sample selection priority of severe occluded GTs. In addition, we reveal the deficiency of currently popular post-processing algorithms, Non-Maximum Suppression (NMS) and Soft-NMS for dense pedestrian detection, and propose Hierarchical-NMS (H-NMS) whose strategy diverges with different Intersection over Union (IoU) levels. Experimental results on Crowdhuman, the largest dense pedestrian detection dataset at present, show our approach optimizes the Average Precision (AP) and Miss Rate (MR) of Fully Convolutional One Stage Object Detection (FCOS) by 4.05% and 5.35% and outperforms many state-of-the-art detectors, respectively. Besides, our method does not bring any increase in calculation during inference.

Multi-view pedestrian detection method based on multi-view Bayesian network

Robust Multiple Cameras Pedestrian Detection with Multi-View Bayesian Network.

Towards Accurate Dense Pedestrian Detection Via Occlusion-Prediction Aware Label Assignment and Hierarchical-Nms.

A Novel Approach to Design the Fast Pedestrian Detection for Video Surveillance System

Pedestrian Detection with Multi-View Convolution Fusion Algorithm

MULTI-STREAM REGION PROPOSAL NETWORK FOR PEDESTRIAN DETECTION

Vision Based Multi-pedestrian Tracking Using Adaptive Detection and Clustering.

A Part-Aware Multi-Scale Fully Convolutional Network for Pedestrian Detection

MAPD: An improved multi-attribute pedestrian detection in a crowd

Multi-pose Pedestrian Detection Based on Posterior Multiple Sparse Dictionaries

Multi-scale Feature Balance Enhancement Network for Pedestrian Detection

A Pedestrian-Detection Method Based on Heterogeneous Features and Ensemble of Multi-View–Pose Parts

Multi-Pedestrian Detection from Effective Proposal in Crowd Scene

Pedestrian detection by multiple decision-based neural networks

Coupled Network for Robust Pedestrian Detection With Gated Multi-Layer Feature Extraction and Deformable Occlusion Handling

Joint HFaster-RCNN and Bayesian Posterior Probabilities for Pedestrian Detection

Multi‐scale Pedestrian Detection with Global–local Attention and Multi‐scale Receptive Field Context

Multi-Attribute NMS: an Enhanced Non-Maximum Suppression Algorithm for Pedestrian Detection in Crowded Scenes

Multi Features Combination for Pedestrian Detection.

Multi-branch detection network based on trigger attention for pedestrian detection under occlusion

Multi-view and Multi-Plane Data Fusion for Effective Pedestrian Detection in Intelligent Visual Surveillance