Non-destructive classification of sturgeon stress using cross-modal data fusion and multi-input deep learning models

Wentao Huang, Yangfeng Wang, Jie Xia, Xinyi Jin, Hongliang Zhu, Branko Glamuzina, Wenyong Yu, Xiaoshuan Zhang

Abstract: Highlights•A sturgeon stress classification technique based on flexible sensing and data fusion algorithms.•A multi-channel data fusion algorithm based on Alexnet and ANN.•A cross-modal stress classification network (CM-SCN) model based on physiological, environmental, and visual features.•The CM-SCN model outperforms a visual CNN model and a physiological-environmental ANN model.•The CM-SCN model provides an effective and reliable method for monitoring sturgeon health.