Water quality parameter analysis model based on fish behavior

Abstract

In the process of aquaculture, the quality of water environment directly determines the quality of aquaculture. Deterioration of water quality will directly cause a decline in aquaculture production, and in severe cases, it will cause a large number of deaths of aquatic organisms and serious economic losses to aquaculture enterprises. Therefore, it is important to real-time monitor water quality parameters in aquaculture. In this paper, a water quality monitoring model based on fish behavior is proposed with the oplegnathus punctatus as the research object. We do not need to install complicated equipment, and only need the image data captured by the camera to accomplish the real-time monitoring of water quality parameters. In terms of model, we also added the CBAM attention mechanism, residual module and a series of training strategies, so that the model can focus on the valuable fish behavior information in the image while focusing on the global information. The experimental results show that the accuracy of the model in the validation set can reach 97.2%, and the inference speed can reach 144.36 FPS. And the experiment shows that our model has good generalization performance.