论文标题
均衡损失V2：长尾对象检测的新梯度平衡方法
Equalization Loss v2: A New Gradient Balance Approach for Long-tailed Object Detection
论文作者
论文摘要
最近提出的解耦训练方法是长尾对象检测的主要范式。但是它们需要一个额外的微调阶段，并且表示形式和分类器的分离优化可能会导致次优结果。但是，端到端训练方法（例如均衡损失（EQL））的表现仍然比脱钩的训练方法差。在本文中，我们揭示了长尾对象检测中的主要问题是阳性和负面因素之间的不平衡梯度，并发现EQL无法很好地解决它。为了解决不平衡梯度的问题，我们引入了一种新版本的均衡损失，称为均衡损失v2（eql v2），这是一种新颖的梯度引导重新升级机制，可以独立且平均地重新平衡每个类别的训练过程。在具有挑战性的LVIS基准上进行了广泛的实验。 EQL V2优于原产质EQL的总AP约为4分，在罕见类别上有14-18分提高了。更重要的是，它还超过了脱钩的训练方法。 EQL V2无需进一步调整开放图像数据集，将EQL提高了7.3点AP，显示出强大的概括能力。代码已在https://github.com/tztztztztztz/eqlv2上发布
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
