Application of large block coal recognition method based on perspective transformation and SNc-YOLOv5
To enhance the precision of identifying large coal chunks in belt conveyors,mitigate belt wear from coal accumulation and increase the longevity of belt conveyors,a novel method rooted in per-spective transformation and SNc-YOLOv5 was proposed.Firstly,the perspective transformation was uti-lized to refine the original image,successfully omitting areas unrelated to the conveyor and adjusting the conveyor's coordinates.Then,the SNc-YOLOv5 model was undertaken to establish deep learning model on these standardized images,yielding a model adept at recognizing large block coal.The efficacy of this approach was confirmed through both test analysis and on-site application.The results show that the precision rate on the dataset of Mine 1 is 94.8％,with a recall rate of 83.2％,while on Mine 2's dataset,the precision rate is 92.8％,with a recall rate of 85.3％.In field applications,the confidence level reaches 0.9,compared with alternative techniques,this method outperforms in terms of both accu-racy and recall.By focusing on the conveyor region during image preprocessing,the method not only standardizes the image but also elevates the precision of large block coal identification,which thus con-tinuous real-time surveillance,fortifying the safety and longevity of belt conveyors.
large block coal identificationbelt conveyorperspective transformationnormalisation
万方数据
维普
