Machine Learning Sorting Method of Bauxite Based on SE-Enhanced Network

A fast and accurate bauxite recognition method combining an attention module a clustering algorithm is proposed in this paper. By introducing the K-means into YOLOv4 network embedding SE module, we calculate corresponding anchor box value, enhance feature learning ability of to bauxite, automatically learn importance different channel features, improve accuracy target detection. In experiment, 2189 photos were taken screened as detection datasets, targets divided four categories: No. 55, 65, 70, Nos. 72–73. selecting category volume balanced optimal model was obtained after training 7000 times, so that average sorting reached 99%, reasoning speed better than 0.05 s. Realizing high-speed high-precision greatly improves mining efficiency industry. At same time, provides key technical support for practical application type engineering.