Research on Underwater Image Semantic Segmentation Method Based on SegNet
Jialun Wang, Fusheng Zha, Xiuwen Bi · 2022
At present, there is little research on underwater image semantic segmentation, and the effect of training using the traditional semantic segmentation network model on land is not ideal.This paper analyzes the current development of underwater image semantic segmentation, establishes a multi-objective underwater image semantic segmentation dataset, and proposes a network model for the impact of underwater complex conditions on semantic segmentation.Experiments show that this network has better segmentation effect than SegNet network in processing underwater images.