The Application of Semantic Segmentation on 2D images

Abstract

A fundamental problem in computer vision is semantic segmentation, which calls for the algorithm to categorize each pixel in the picture and provide the precise details of the category. Semantic segmentation is being employed extensively in a variety of applications, including autonomous vehicles and medical imaging. An overview of similar semantic segmentation approaches is given in this study. First, this paper gives a brief overview of the history and vocabulary of semantic segmentation. The key datasets for semantic segmentation, conventional segmentation models, and fundamental deep learning techniques for semantic segmentation will then be covered. In particular, traditional methods centered on context models and deep learning-centered methods are discussed in detail. Finally, we review several assessment techniques, including their benefits and drawbacks, and outline the key issues facing semantic segmentation today. In addition, this study seeks to provide an overview of the relevant literature and the difficulties in semantic segmentation. Finally, the paper summarizes the semantic segmentation and prospects the future.