Background Noise Reduction of Attention Map for Weakly Supervised Semantic Segmentation
This work addresses background noise contamination in weakly supervised semantic segmentation, which is an incremental improvement over the existing TransCAM method.
The paper tackles the problem of background noise in attention maps for weakly supervised semantic segmentation, proposing a method that reduces this noise and improves pseudo label accuracy, achieving segmentation performances of 70.5% on PASCAL VOC 2012 validation, 71.1% on test data, and 45.9% on MS COCO 2014.
In weakly-supervised semantic segmentation (WSSS) using only image-level class labels, a problem with CNN-based Class Activation Maps (CAM) is that they tend to activate the most discriminative local regions of objects. On the other hand, methods based on Transformers learn global features but suffer from the issue of background noise contamination. This paper focuses on addressing the issue of background noise in attention weights within the existing WSSS method based on Conformer, known as TransCAM. The proposed method successfully reduces background noise, leading to improved accuracy of pseudo labels. Experimental results demonstrate that our model achieves segmentation performance of 70.5% on the PASCAL VOC 2012 validation data, 71.1% on the test data, and 45.9% on MS COCO 2014 data, outperforming TransCAM in terms of segmentation performance.