Construction method of adhesive tape surface defect data based on conditional GAN and adaptive fusion
In view of the practical problem that surface defects are prone to occur in coal mine belt conveyor in complex working environment,which affects production and safety,the image synthesis algorithm based on conditional adversity-generating network and traditional image processing technology are adopted to improve the accuracy and stability of defect detection.The research results show that the adaptive fusion algorithm can generate real and stable composite images,and the generated samples do not need to manually mark the target box,which helps to quickly establish the training sample set required for supervised learning,and solves the challenge of the algorithm in the coal mine environment.Using the data synthesis algorithm,the target detection rate is increased by more than 5%.
belt conveyorGANadaptive fusionsurface defectstarget detection rate
NETL
NSTL
万方数据
