煤矸石分拣是煤炭开采和加工过程中的重要环节,可以有效降低煤炭后期加工难度和成本。针对目前我国煤矸石分拣工艺复杂、分拣效率低、人工选矸工作难度大等问题,提出了一种基于深度学习的煤矸石检测方法。该方法采用YOLOv7深度学习算法为核心,通过制作煤矸石数据集、训练检测模型、搭建煤矸石检测平台,实现了煤矸石实时智能分拣。试验结果表明,YOLOv7模型的mAP为96.70%,检测速度为69fps,相比于YOLOv5、SSD和Faster-RCNN算法具有明显优势。

Coal gangue sorting is an important link in the coal mining and processing process, which caneffectively reduce the difficulty and cost of coal processing in the later stage. A deep learning based coal ganguedetection method was proposed to address the problems of complex coal gangue sorting processes, low sortingefficiency, and high difficulty in manual selection in China. This method adopts the YOLOv7 deep learningalgorithm as the core, and achieves real-time intelligent sorting of coal gangue by creating a coal gangue dataset,training detection models, and building a coal gangue detection platform. The experimental results showed that themAP of YOLOv7 model was 96. 70%, and the detection speed was 69fps, which had significant advantagescompared to YOLOv5, SSD, and Faster-RCNN algorithms.