Alert button

Non-Salient Region Object Mining for Weakly Supervised Semantic Segmentation

Add code
Alert button
Mar 26, 2021
Figure 1 for Non-Salient Region Object Mining for Weakly Supervised Semantic Segmentation
Figure 2 for Non-Salient Region Object Mining for Weakly Supervised Semantic Segmentation
Figure 3 for Non-Salient Region Object Mining for Weakly Supervised Semantic Segmentation
Figure 4 for Non-Salient Region Object Mining for Weakly Supervised Semantic Segmentation

Share this with someone who'll enjoy it:

Semantic segmentation aims to classify every pixel of an input image. Considering the difficulty of acquiring dense labels, researchers have recently been resorting to weak labels to alleviate the annotation burden of segmentation. However, existing works mainly concentrate on expanding the seed of pseudo labels within the image's salient region. In this work, we propose a non-salient region object mining approach for weakly supervised semantic segmentation. We introduce a graph-based global reasoning unit to strengthen the classification network's ability to capture global relations among disjoint and distant regions. This helps the network activate the object features outside the salient area. To further mine the non-salient region objects, we propose to exert the segmentation network's self-correction ability. Specifically, a potential object mining module is proposed to reduce the false-negative rate in pseudo labels. Moreover, we propose a non-salient region masking module for complex images to generate masked pseudo labels. Our non-salient region masking module helps further discover the objects in the non-salient region. Extensive experiments on the PASCAL VOC dataset demonstrate state-of-the-art results compared to current methods.

* accepted by IEEE Conference on Computer Vision and Pattern Recognition, 2021  

Share this with someone who'll enjoy it: