Ore Particle Size Detection Based on Improved YOLOv10 Algorithm

The realization of ore size statistics through machine vision plays a vital role in the safe and efficient production of mining industry. However, the proportion of small grains of ore in the ore conveyor belt is high, and the ore blocks each other a lot, resulting in the low accuracy of the existing target detection model, and it is difficult to meet the accuracy requirements of particle size detection. In order to improve the accuracy of the target detection model in the ore size statistics task, this paper proposes an improved YOLOV10 method. Firstly, PIoU is used as the loss function, which improves the ability of the model to focus on anchor boxes with moderate quality. Secondly, SKNet is introduced to optimize the backbone network, which improves the model's attention to key information. Finally, a feature processing module named Snake-Multi-Scale Block(SMSB) is used to enhance the ability of the model to extract features. The experimental results show that the precision of the improved model is increased by 4.1%, the recall rate is increased by 5.0%, mAP50 is increased by 5.0%, and mAP50: 90 is increased by 3.9%. It is proved that the new method can effectively improve the recognition accuracy of ore size.