ICFRNet: Image Complexity Prior Guided Feature Refinement for Real-time Semantic Segmentation

• DOI: 10.48550/arxiv.2408.13771
• Publication Date: 2024-08-25
• Authors: Xin Zhang; Teodor Boyadzhiev; Jinglei Shi; Jufeng Yang
• Keywords: Feature (linguistics); Semantic feature

In this paper, we leverage image complexity as a prior for refining segmentation features to achieve accurate real-time semantic segmentation. The design philosophy is based on the observation that different pixel regions within an exhibit varying levels of complexity, with higher complexities posing greater challenge We thus introduce guidance and propose Image Complexity prior-guided Feature Refinement Network (ICFRNet). This network aggregates both produce attention map Guided Attention (ICGA) module. optimize in terms prediction tasks combined loss function. Experimental results Cityscapes CamViD datasets have shown our ICFRNet achieves accuracy competitive efficiency