Towards Long-Range Pixels Connection for Context-Aware Semantic Segmentation

Abstract

Semantic segmentation is one of the challenging tasks in computer vision. Before the advent of deep learning, hand-crafted features were used to semantically extract the region-of-interest (ROI). Deep learning has recently achieved enormous response in semantic image segmentation. The previously developed U-Net inspired architectures operate with continuous stride and pooling operations, leading to spatial data loss. Also, the methods lack establishing long-term pixels connection to preserve context knowledge and reduce spatial loss in prediction. This article developed encoder-decoder architecture with a sequential block embedded in long skip-connections and densely connected convolution blocks. The network non-linearly combines the feature maps across encoder-decoder paths for finding dependency and correlation between image pixels. Additionally, the densely connected convolutional blocks are kept in the final encoding layer to reuse features and prevent redundant data sharing. The method applied batch-normalization to reduce internal covariate shift in data distributions. We have used LUNA, ISIC2018, and DRIVE datasets to reflect three different segmentation problems (lung nodules, skin lesions, vessels) and claim the effectiveness of the proposed architecture. The network is also compared with other techniques designed to highlight similar problems. It is found through empirical evidence that our method shows promising results when compared with other segmentation techniques.