Deep Learning Based Mineral Image Classification Combined With Visual Attention Mechanism

Mineral image classification technology based on machine vision is an efficient system for ore sorting. With the development of artificial intelligence and computer technology, deep learning-based mineral gradually applied to However, there a bottleneck in improving accuracy, feature extraction ability CNNs model relatively limited multi-category tasks. Therefore, four visual attention blocks are designed embedded existing model, new models mechanism proposed. Then, referring building strategies different depth ResNet, we build various embedding with visualize by Grad-CAM observe change weight distributions values. Finally, using confusion matrices, this experiment systematically evaluates performance proposed analyzes misjudgment rate.