VALSE 论文速览 第96期：标签匹配的半监督目标检测方法

论文题目：Label Matching Semi-Supervised Object Detection

作者列表：陈斌斌 (海康威视研究院)，陈伟杰 (海康威视研究院，浙江大学)，杨世才 (海康威视研究院)，禤韵怡 (海康威视研究院)，宋杰 (浙江大学)，谢迪 (海康威视研究院)，浦世亮 (海康威视研究院)，宋明黎 (浙江大学)，庄越挺 (浙江大学)

B站观看网址：

https://www.bilibili.com/video/BV1eB4y137vg/

论文摘要：

Semi-supervised object detection has made significant progress with the development of mean teacher driven self-training. Despite the promising results, the label mismatch problem is not yet fully explored in the previous works, leading to severe confirmation bias during self-training. In this paper, we delve into this problem and propose a simple yet effective LabelMatch framework from two different yet complementary perspectives, i.e., distribution-level and instance-level. For the former one, it is reasonable to approximate the class distribution of the unlabeled data from that of the labeled data according to Monte Carlo Sampling. Guided by this weakly supervision cue, we introduce a re-distribution mean teacher, which leverages adaptive label-distribution-aware confidence thresholds to generate unbiased pseudo labels to drive student learning. For the latter one, there exists an overlooked label assignment ambiguity problem across teacher-student models. To remedy this issue, we present a novel label assignment mechanism for self-training framework, namely proposal self-assignment, which injects the proposals from student into teacher and generates accurate pseudo labels to match each proposal in the student model accordingly. Experiments on both MS-COCO and PASCAL-VOC datasets demonstrate the considerable superiority of our proposed framework to other state-of-the-arts.

论文信息：

[1] Binbin Chen, Weijie Chen, Shicai Yang, et al. Label Matching Semi-Supervised Object Detection. CVPR 2022.

论文链接：

[Label Matching Semi-Supervised Object Detection (thecvf.com)]

[Chen_Label_Matching_Semi-Supervised_CVPR_2022_supplemental.pdf (thecvf.com)]

代码链接：

[https://github.com/hikvision-research/SSOD]

视频讲者简介：

陈斌斌，海康威视研究院算法工程师，毕业于华中科技大学人工智能学院。主要研究方向为目标检测，半监督目标检测和多模态等。

特别鸣谢本次论文速览主要组织者：

月度轮值AC：冯尊磊 (浙江大学)，徐易 (大连理工大学)

季度责任AC：张姗姗 (南京理工大学)