Adaptive Early-Learning Correction for Segmentation from Noisy Annotations
Adaptive Early-Learning Correction for Segmentation from Noisy Annotations
复制标题
DOI：
10.1109/cvpr52688.2022.00263
复制
发表时间：
2021-10
期刊：
影响因子：
--
通讯作者：
Sheng Liu;Kangning Liu;Weicheng Zhu;Yiqiu Shen;C. Fernandez‐Granda
中科院分区：
文献类型：
--
作者：
Sheng Liu;Kangning Liu;Weicheng Zhu;Yiqiu Shen;C. Fernandez‐Granda
存在噪声注释的深度学习在分类中得到了广泛的研究，但在分割任务中研究得很少。在这项工作中，我们研究了在不准确的注释数据上训练的深度分割网络的学习动态。我们观察到一个以前在分类背景下报道过的现象：网络倾向于在“早期学习”阶段首先拟合干净的像素级标签，然后最终记住错误的注释。然而，与分类相反，分段记忆并不同时出现在所有的语义类别中。受这些发现的启发，我们提出了一种新的方法，从噪声注释分割两个关键要素。首先，我们在训练过程中分别检测每个类别的记忆阶段的开始。这使我们能够自适应地纠正嘈杂的注释，以利用早期学习。其次，我们引入了一个正则化项，该项可以强制跨尺度的一致性，以提高对注释噪声的鲁棒性。我们的方法优于标准方法的医学成像分割任务，噪声合成模仿人类注释错误。它还为弱监督语义分割中存在的真实噪声注释提供了鲁棒性，在PASCAL VOC 2012上实现了最先进的结果。11代码可在https://github.com/Kangningthu/ADELE上获得
Deep learning in the presence of noisy annotations has been studied extensively in classification, but much less in segmentation tasks. In this work, we study the learning dynamics of deep segmentation networks trained on inaccurately annotated data. We observe a phenomenon that has been previously reported in the context of classification: the networks tend to first fit the clean pixel-level labels during an “early-learning” phase, before eventually memorizing the false annotations. However, in contrast to classification, memorization in segmentation does not arise simultaneously for all semantic categories. Inspired by these findings, we propose a new method for segmentation from noisy annotations with two key elements. First, we detect the beginning of the memorization phase separately for each category during training. This allows us to adaptively correct the noisy annotations in order to exploit early learning. Second, we incorporate a regularization term that enforces consistency across scales to boost robustness against annotation noise. Our method outperforms standard approaches on a medical-imaging segmentation task where noises are synthesized to mimic human annotation errors. It also provides robustness to realistic noisy annotations present in weakly-supervised semantic segmentation, achieving state-of-the-art results on PASCAL VOC 2012. 11Code is available at https://github.com/Kangningthu/ADELE