Recent progress in semantic image segmentation

TL;DR

This paper reviews recent advances in semantic image segmentation, emphasizing deep neural network approaches, datasets, and various methodological innovations, highlighting significant progress and ongoing challenges in the field.

Contribution

It provides a comprehensive survey of traditional and deep learning-based segmentation methods, categorizing recent DNN techniques and analyzing their key aspects and developments.

Findings

Deep neural networks have significantly advanced semantic segmentation.

Various DNN architectures like FCN, dilated convolutions, and pyramid methods improve accuracy.

Benchmark datasets have facilitated progress and evaluation in the field.

Abstract

Semantic image segmentation, which becomes one of the key applications in image processing and computer vision domain, has been used in multiple domains such as medical area and intelligent transportation. Lots of benchmark datasets are released for researchers to verify their algorithms. Semantic segmentation has been studied for many years. Since the emergence of Deep Neural Network (DNN), segmentation has made a tremendous progress. In this paper, we divide semantic image segmentation methods into two categories: traditional and recent DNN method. Firstly, we briefly summarize the traditional method as well as datasets released for segmentation, then we comprehensively investigate recent methods based on DNN which are described in the eight aspects: fully convolutional network, upsample ways, FCN joint with CRF methods, dilated convolution approaches, progresses in backbone network, pyramid methods, Multi-level feature and multi-stage method, supervised, weakly-supervised and unsupervised methods. Finally, a conclusion in this area is drawn.