Research Article
Feature Fusion in Deep-Learning Semantic Image Segmentation: A Survey
@INPROCEEDINGS{10.1007/978-3-031-06371-8_18, author={Jie Yuan and Zhaoyi Shi and Shuo Chen}, title={Feature Fusion in Deep-Learning Semantic Image Segmentation: A Survey}, proceedings={Science and Technologies for Smart Cities. 7th EAI International Conference, SmartCity360°, Virtual Event, December 2-4, 2021, Proceedings}, proceedings_a={SMARTCITY}, year={2022}, month={6}, keywords={Feature fusion Deep learning Semantic segmentation}, doi={10.1007/978-3-031-06371-8_18} }- Jie Yuan
Zhaoyi Shi
Shuo Chen
Year: 2022
Feature Fusion in Deep-Learning Semantic Image Segmentation: A Survey
SMARTCITY
Springer
DOI: 10.1007/978-3-031-06371-8_18
Abstract
Semantic image segmentation is a necessary research and application direction for intelligent systems. Many researchers have tried to design advanced feature fusion to extract beneficial information from different feature maps selectively. However, there is no published review currently that focuses on feature fusion for semantic image segmentation. Therefore, we seek to compile related works and analyze the trends and challenges of feature fusion. In this paper, we introduce feature fusion modules based on different semantic image segmentation models. Then, we analyze typical and state-of-the-art approaches in terms of several effective from fusion. Third, we comprehensively present fusion strategies. Finally, we summarize the challenges as well as the development trend of feature fusion. This survey infers that although significant developments have been obtained, there is still plenty of room for improvement of feature fusion. Interpretability in deep-learning segmentation and the application of novel mechanisms have been important directions for future exploration.
