Attention-guided chained context aggregation for semantic segmentation

The way features propagate in Fully Convolutional Networks is of momentous importance to capture multi-scale contexts for obtaining precise segmentation masks. This paper proposes a novel series-parallel hybrid paradigm called the Chained Context Aggregation Module (CAM) enrich feature representation. CAM gains various spatial scales through chain-connected ladder-style information flows and fuses them two-stage process, namely pre-fusion re-fusion. serial flow continuously increases receptive fields output neurons those parallel encode different region-based contexts. Each shallow encoder-decoder with appropriate down-sampling sufficiently contextual information. We further adopt an attention model guide Based on these developments, we construct Network (CANet), which employs asymmetric decoder recover details prediction maps. conduct extensive experiments six challenging datasets, including Pascal VOC 2012, Context, Cityscapes, CamVid, SUN-RGBD GATECH. Results evidence that CANet achieves state-of-the-art or competitive performance.