Title:An Abstraction Model for Semantic Segmentation Algorithms

Abstract:Semantic segmentation classifies each pixel in the image. Due to its advantages, semantic segmentation is used in many tasks, such as cancer detection, robot-assisted surgery, satellite image analysis, and self-driving cars. Accuracy and efficiency are the two crucial goals for this purpose, and several state-of-the-art neural networks exist. By employing different techniques, new solutions have been presented in each method to increase efficiency and accuracy and reduce costs. However, the diversity of the implemented approaches for semantic segmentation makes it difficult for researchers to achieve a comprehensive view of the field. In this paper, an abstraction model for semantic segmentation offers a comprehensive view of the field. The proposed framework consists of four general blocks that cover the operation of the majority of semantic segmentation methods. We also compare different approaches and analyze each of the four abstraction blocks' importance in each method's operation.