Abstract:Object attention maps generated by image classifiers are usually used as priors for weakly supervised semantic segmentation. However, attention maps usually locate the most discriminative object parts. The lack of integral object localization maps heavily limits the performance of weakly supervised segmentation approaches. This paper attempts to investigate a novel way to identify entire object regions in a weakly supervised manner. We observe that image classifiers' attention maps at different training phases may focus on different parts of the target objects. Based on this observation, we propose an online attention accumulation (OAA) strategy that utilizes the attention maps at different training phases to obtain more integral object regions. Specifically, we maintain a cumulative attention map for each target category in each training image and utilize it to record the discovered object regions at different training phases. Albeit OAA can effectively mine more object regions for most images, for some training images, the range of the attention movement is not large, limiting the generation of integral object attention regions. To overcome this problem, we propose incorporating an attention drop layer into the online attention accumulation process to enlarge the range of attention movement during training explicitly. Our method (OAA) can be plugged into any classification network and progressively accumulate the discriminative regions into cumulative attention maps as the training process goes. Additionally, we also explore utilizing the final cumulative attention maps to serve as the pixel-level supervision, which can further assist the network in discovering more integral object regions. When applying the resulting attention maps to the weakly supervised semantic segmentation task, our approach improves the existing state-of-the-art methods on the PASCAL VOC 2012 segmentation benchmark, achieving a mIoU score of 67.2 percent on the test set.

Mixed High-Order Attention Network for Weakly- Supervised Semantic Segmentation

Online Attention Accumulation for Weakly Supervised Semantic Segmentation

Polarized Attention Weak Supervised Semantic Segmentation Network

Saliency guided deep network for weakly-supervised image segmentation

Lightweight Attention Network for Very High-Resolution Image Semantic Segmentation

Decoupled Spatial Neural Attention for Weakly Supervised Semantic Segmentation.

Collaborative Semantic Segmentation with Image Labels.

Research on Image Semantic Segmentation Based on Hybrid Cascade Feature Fusion and Detailed Attention Mechanism

Semantic Image Segmentation Based On Attentions To Intra Scales And Inner Channels

Point Attention Network for Point Cloud Semantic Segmentation.

Self-Supervised Equivariant Attention Mechanism for Weakly Supervised Semantic Segmentation

Multi-layered self-attention mechanism for weakly supervised semantic segmentation

Enhanced Pseudo-Label Generation with Self-supervised Training for Weakly-supervised Semantic Segmentation

Bottom-Up Top-Down Cues for Weakly-Supervised Semantic Segmentation

Weakly Supervised Semantic Segmentation With Consistency-Constrained Multiclass Attention for Remote Sensing Scenes

Hybrid Multiple Attention Network for Semantic Segmentation in Aerial Images

Partial Class Activation Attention for Semantic Segmentation

Saliency Guided Self-Attention Network for Weakly and Semi-Supervised Semantic Segmentation

Weakly supervised segmentation via instance-aware propagation

Semantic Segmentation With Attention Mechanism for Remote Sensing Images

Expectation-Maximization Attention Networks for Semantic Segmentation