Xingang Wang’s research while affiliated with Institute of Automation, Chinese Academy of Sciences and other places

What is this page?

This page lists works of an author who doesn't have a ResearchGate profile or hasn't added the works to their profile yet. It is automatically generated from public (personal) data to further our legitimate goal of comprehensive and accurate scientific recordkeeping. If you are this author and want this page removed, please let us know.

Publications (1)

Figure 2. The pipeline to obtain the information difference maps, where the Information Perturbation Module (IPM) regulates the information flow by adding noise perturbation. The auxiliary branch is added to generate the comprehensive perturbed maps.
Figure 3. Illustration of different CAMs. (a) The perturbed CAMs generated by our proposed IPM module. (b) The original CAMs generated by the traditional classification networks. (c) The information difference maps where the white arrows represent the direction of the information bottleneck from the original most discriminative regions to the non-discriminative regions of the target objects. (d) The ground truth labels of the images.
Figure 4. The architecture of the proposed information perturbation module. δ i denotes the noise perturbation. α i denotes the attentive map obtained from the multi-level features.
Figure 5. The framework of the proposed method for breakthrough the information bottleneck in weakly supervised semantic segmentation.
Per-class segmentation results on PASCAL VOC2012 val set with DeepLab-v2 [5].

+1

Exploring Information Bottleneck for Weakly Supervised Semantic Segmentation
  • Chapter
  • Full-text available

September 2023

·

30 Reads

Jie Qin

·

Yueming Lyu

·

Xingang Wang

Image-level weakly supervised semantic segmentation (WSSS) has attracted much attention due to the easily acquired class labels. Most existing methods resort to utilizing Class Activation Maps (CAMs) obtained from the classification network to play as the initial pseudo labels. However, the classifiers only focus on the most discriminative regions of the target objects, which is referred to as the information bottleneck from the perspective of the information theory. To alleviate this information bottleneck limitation, we propose an Information Perturbation Module (IPM) to explicitly obtain the information difference maps, which provide the accurate direction and magnitude of the information compression in the classification network. After that, an information bottleneck breakthrough mechanism with three branches is proposed to overcome the information bottleneck in the classification network for segmentation. Additionally, a diversity regularization on the generated two information difference maps is proposed to improve the diversity of the output CAMs. Extensive experiments on PASCAL VOC2012 val and test sets demonstrate that the proposed method can effectively improve the weakly supervised semantic segmentation performance of the advanced approaches.

Download