An underwater target detection method based on context-aware attention

Abstract

The application provides an underwater target detection method based on context-aware attention, comprising the following steps: acquiring an underwater image dataset; preprocessing the image; constructing an underwater target detection model based on context-aware attention, the model comprising a backbone network, a feature fusion network and a detection head network, and an attention module being arranged between the backbone network and the feature fusion network and between the feature fusion network and the detection head network; training the underwater target detection model based on context-aware attention based on the preprocessed image data; inputting a to-be-detected image into the trained underwater target detection model based on context-aware attention, and obtaining the output of the model as a target detection result. The application designs an attention mechanism based on context awareness to improve the target perception ability of the detection network in the underwater environment, and the attention mechanism combined with a YOLOv5 target detection network can realize accurate and efficient underwater target detection.

Images