Equalization Loss v2: A New Gradient Balance Approach for Long-tailed Object Detection

Abstract

Recently proposed decoupled training methods emerge as a dominant paradigm for long-tailed object detection. But they require an extra fine-tuning stage, and the disjointed optimization of representation and classifier might lead to suboptimal results. However, end-to-end training methods, like equalization loss (EQL), still perform worse than decoupled training methods. In this paper, we reveal the main issue in long-tailed object detection is the imbalanced gradients between positives and negatives, and find that EQL does not solve it well. To address the problem of imbalanced gradients, we introduce a new version of equalization loss, called equalization loss v2 (EQL v2), a novel gradient guided reweighing mechanism that re-balances the training process for each category independently and equally. Extensive experiments are performed on the challenging LVIS benchmark. EQL v2 outperforms origin EQL by about 4 points overall AP with 14-18 points improvements on the rare categories. More importantly, it also surpasses decoupled training methods. Without further tuning for the Open Images dataset, EQL v2 improves EQL by 7.3 points AP, showing strong generalization ability. Codes have been released at https://github.com/tztztztztz/eqlv2

CVPR2021, codes: https://github.com/tztztztztz/eqlv2

References in corpus (9)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• MMDetection: Open MMLab Detection Toolbox and Benchmark
• The Open Images Dataset V4: Unified image classification, object detection, and visual relationship detection at scale
• Meta-Weight-Net: Learning an Explicit Mapping For Sample Weighting
• Decoupling Representation and Classifier for Long-Tailed Recognition
• Long-Tailed Classification by Keeping the Good and Removing the Bad Momentum Causal Effect
• Feature Space Augmentation for Long-Tailed Data
• Forest R-CNN: Large-Vocabulary Long-Tailed Object Detection and Instance Segmentation
• The Devil is in Classification: A Simple Framework for Long-tail Object Detection and Instance Segmentation

Cited by in corpus (3)

• ACE: Ally Complementary Experts for Solving Long-Tailed Recognition in One-Shot
• Forest R-CNN: Large-Vocabulary Long-Tailed Object Detection and Instance Segmentation
• Disentangle Your Dense Object Detector