AutoAssign: Differentiable Label Assignment for Dense Object Detection

Abstract

Determining positive/negative samples for object detection is known as label assignment. Here we present an anchor-free detector named AutoAssign. It requires little human knowledge and achieves appearance-aware through a fully differentiable weighting mechanism. During training, to both satisfy the prior distribution of data and adapt to category characteristics, we present Center Weighting to adjust the category-specific prior distributions. To adapt to object appearances, Confidence Weighting is proposed to adjust the specific assign strategy of each instance. The two weighting modules are then combined to generate positive and negative weights to adjust each location's confidence. Extensive experiments on the MS COCO show that our method steadily surpasses other best sampling strategies by large margins with various backbones. Moreover, our best model achieves 52.1% AP, outperforming all existing one-stage detectors. Besides, experiments on other datasets, e.g., PASCAL VOC, Objects365, and WiderFace, demonstrate the broad applicability of AutoAssign.

arXiv commentRevised for submission

References in corpus (10)

• Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
• YOLOv3: An Incremental Improvement
• DenseBox: Unifying Landmark Localization with End to End Object Detection
• FreeAnchor: Learning to Match Anchors for Visual Object Detection
• CornerNet: Detecting Objects as Paired Keypoints
• RepPoints: Point Set Representation for Object Detection
• Bottom-up Object Detection by Grouping Extreme and Center Points
• MetaAnchor: Learning to Detect Objects with Customized Anchors
• Soft Anchor-Point Object Detection
• Learning from Noisy Anchors for One-stage Object Detection

Cited by in corpus (27)

• YOLOX: Exceeding YOLO Series in 2021
• You Only Look at One Sequence: Rethinking Transformer in Vision through Object Detection
• Probabilistic two-stage detection
• Efficient DETR: Improving End-to-End Object Detector with Dense Prior
• A Gated Cross-domain Collaborative Network for Underwater Object Detection
• ByteTrack: Multi-Object Tracking by Associating Every Detection Box
• Domain generalization in deep learning-based mass detection in mammography: A large-scale multi-center study
• FCOS: A simple and strong anchor-free object detector
• Dynamic Label Assignment for Object Detection by Combining Predicted IoUs and Anchor IoUs
• Task-wise Sampling Convolutions for Arbitrary-Oriented Object Detection in Aerial Images
• EARL: An Elliptical Distribution aided Adaptive Rotation Label Assignment for Oriented Object Detection in Remote Sensing Images
• Generalized Focal Loss V2: Learning Reliable Localization Quality Estimation for Dense Object Detection
• Implicit Feature Pyramid Network for Object Detection
• End-to-End Object Detection with Fully Convolutional Network
• LLA: Loss-aware Label Assignment for Dense Pedestrian Detection
• OTA: Optimal Transport Assignment for Object Detection
• You Only Look One-level Feature
• Localization Distillation for Dense Object Detection
• Pseudo-IoU: Improving Label Assignment in Anchor-Free Object Detection
• IQDet: Instance-wise Quality Distribution Sampling for Object Detection
• TE-YOLOF: Tiny and efficient YOLOF for blood cell detection
• To Make Yourself Invisible with Adversarial Semantic Contours
• Progressive Hard-case Mining across Pyramid Levels for Object Detection
• Higher Performance Visual Tracking with Dual-Modal Localization
• Modulating Localization and Classification for Harmonized Object Detection
• Bootstrap Your Object Detector via Mixed Training
• Learning Oriented Remote Sensing Object Detection via Naive Geometric Computing