Underground Mine Road Detection Using Deep Learning Technique

Semantic segmentation of underground mine roads is very important to efficiently obtain road information from images. The boundary not obvious, the environment complex, and identification difficult. In order effectively realize accurate roads, a network model using deep learning technique proposed. Choosing BiSeNet as basic framework, adopting unified attention fusion module, channel spatial enrich feature representation can reduce loss information. addition, lightweight STDC integrated into backbone computational complexity. Finally, experiments were carried out on roads. experimental results show that mean intersection over union pixel accuracy proposed method reached 89.34% 98.34%, respectively, recognition speed 23 f/s when identifying this study, trained by technology solve problem with high accuracy.