Research on defective cotton foreign fiber detection algorithm based on lightweight YOLOv8
To solve the problems of high difficulty and poor real-time detection of small target heterofiber in defective cotton,this paper proposes a defect detection algorithm for foreign fibers in defective cotton based on a lightweight YOLOv8 model(YOLOv8-MMW).Initially,MobileNetV3 lightweight network is introduced to replace the original backbone network to reduce the number of parameters and computational complexity,and improve the real-time detection of network algorithms.Subsequently,a Multi-dimensional Coopera-tive Attention(MCA)mechanism is integrated into the backbone network to boost the interaction of multi-dimensional features,focu-sing the model on small target foreign fibers in defective cotton.Lastly,a dynamic non-monotonic focusing mechanism,WIoUv3,is in-troduced to improve the model's convergence speed and accuracy,enhancing its ability to localize small target foreign fibers.The re-sults show that the improved YOLOv8-MMW model accuracy and average accuracy mean values are 95.5％and 95.8％,respectively.Compared with the original baseline model YOLOv8,the accuracy and average accuracy are increased by 0.9％and 0.1％respective-ly,the model weight is reduced by 56.7％,and the frame rate reaches 367.8 frames/s.The improved model can more quickly and ac-curately identify defective cotton small target foreign fibers,providing technical support for intelligent sorting of defective cotton.
万方数据
维普
