3 个月前

保守-进步协同学习在半监督语义分割中的应用

保守-进步协同学习在半监督语义分割中的应用

摘要

伪监督(pseudo supervision)被视为半监督语义分割中的核心思想,而如何在仅使用高质量伪标签与充分利用所有伪标签之间取得平衡,始终是一个关键的权衡问题。针对这一挑战,本文提出一种新颖的学习范式——保守-渐进协同学习(Conservative-Progressive Collaborative Learning, CPCL)。在该方法中,两个预测网络并行训练,伪监督机制基于两个网络预测结果的一致性与差异性进行构建。其中一个网络通过交集监督(intersection supervision)寻求共性,仅接受高质量伪标签的监督,以确保监督信号的可靠性;另一个网络则通过并集监督(union supervision)保留差异性,接受全部伪标签的监督,以维持探索过程中的好奇心与多样性。由此,实现了保守演化(conservative evolution)与渐进探索(progressive exploration)的协同。为进一步降低可疑伪标签带来的负面影响,损失函数根据预测置信度动态重加权。大量实验结果表明,CPCL在半监督语义分割任务上达到了当前最优(state-of-the-art)性能。

代码仓库

leofansq/CPCL
官方
pytorch

基准测试

基准方法指标
semi-supervised-semantic-segmentation-on-1CPCL (DeepLab v3+ with ResNet-50)
Validation mIoU: 76.98%
semi-supervised-semantic-segmentation-on-15CPCL (DeepLab v3+ with ResNet-101)
Validation mIoU: 77.67%
semi-supervised-semantic-segmentation-on-15CPCL (DeepLab v3+ with ResNet-50)
Validation mIoU: 75.3%
semi-supervised-semantic-segmentation-on-2CPCL (DeepLab v3+ with ResNet-50)
Validation mIoU: 74.6%
semi-supervised-semantic-segmentation-on-21CPCL (DeepLab v3+ with ResNet-101)
Validation mIoU: 73.44
semi-supervised-semantic-segmentation-on-21CPCL (DeepLab v3+ with ResNet-50)
Validation mIoU: 71.66
semi-supervised-semantic-segmentation-on-22CPCL (DeepLab v3+ with ResNet-50)
Validation mIoU: 69.92%
semi-supervised-semantic-segmentation-on-27CPCL (DeepLab v3+ with ResNet-50)
Validation mIoU: 61.88
semi-supervised-semantic-segmentation-on-28CPCL (DeepLab v3+ with ResNet-50)
Validation mIoU: 67.02
semi-supervised-semantic-segmentation-on-29CPCL (DeepLab v3+ with ResNet-50)
Validation mIoU: 72.14
semi-supervised-semantic-segmentation-on-30CPCL (DeepLab v3+ with ResNet-50)
Validation mIoU: 74.25
semi-supervised-semantic-segmentation-on-4CPCL (DeepLab v3+ with ResNet-50)
Validation mIoU: 73.74%
semi-supervised-semantic-segmentation-on-4CPCL (DeepLab v3+ with ResNet-101)
Validation mIoU: 76.4%
semi-supervised-semantic-segmentation-on-8CPCL (DeepLab v3+ with ResNet-50)
Validation mIoU: 78.17%
semi-supervised-semantic-segmentation-on-9CPCL (DeepLab v3+ with ResNet-50)
Validation mIoU: 74.58
semi-supervised-semantic-segmentation-on-9CPCL (DeepLab v3+ with ResNet-101)
Validation mIoU: 77.16

用 AI 构建 AI

从想法到上线——通过免费 AI 协同编程、开箱即用的环境和市场最优价格的 GPU 加速您的 AI 开发

AI 协同编程
即用型 GPU
最优价格
立即开始

Hyper Newsletters

订阅我们的最新资讯
我们会在北京时间 每周一的上午九点 向您的邮箱投递本周内的最新更新
邮件发送服务由 MailChimp 提供
保守-进步协同学习在半监督语义分割中的应用 | 论文 | HyperAI超神经