Learning non-maximum suppression

Abstract

Object detectors have hugely profited from moving towards an end-to-end learning paradigm: proposals, features, and the classifier becoming one neural network improved results two-fold on general object detection. One indispensable component is non-maximum suppression (NMS), a post-processing algorithm responsible for merging all detections that belong to the same object. The de facto standard NMS algorithm is still fully hand-crafted, suspiciously simple, and -- being based on greedy clustering with a fixed distance threshold -- forces a trade-off between recall and precision. We propose a new network architecture designed to perform NMS, using only boxes and their score. We report experiments for person detection on PETS and for general object categories on the COCO dataset. Our approach shows promise providing improved localization and occlusion handling.

Added "Supplementary material" title

References in corpus (9)

• Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
• OverFeat: Integrated Recognition, Localization and Detection using Convolutional Networks
• Learning Convolutional Neural Networks for Graphs
• Rich feature hierarchies for accurate object detection and semantic segmentation
• Convolutional Feature Masking for Joint Object and Stuff Segmentation
• End-to-end people detection in crowded scenes
• End-to-end training of object class detectors for mean average precision
• Spatial Semantic Regularisation for Large Scale Object Detection
• End-to-End Integration of a Convolutional Network, Deformable Parts Model and Non-Maximum Suppression

Cited by in corpus (10)

• Learning to Count Objects in Natural Images for Visual Question Answering
• Daedalus: Breaking Non-Maximum Suppression in Object Detection via Adversarial Examples
• Multi-Task Deep Networks for Depth-Based 6D Object Pose and Joint Registration in Crowd Scenarios
• Visibility Guided NMS: Efficient Boosting of Amodal Object Detection in Crowded Traffic Scenes
• Deep-CEE I: Fishing for Galaxy Clusters with Deep Neural Nets
• Learning Region Features for Object Detection
• C^3Net: End-to-End deep learning for efficient real-time visual active camera control
• Determinantal Point Process as an alternative to NMS
• Fusion of an Ensemble of Augmented Image Detectors for Robust Object Detection
• Proactive Network Maintenance using Fast, Accurate Anomaly Localization and Classification on 1-D Data Series