Broken Rice Detection Dataset Based on Improved YOLOv5
Broken rice detection is an important part of evaluating the quality of rice.Traditional broken rice de-tection is manually selected,time-consuming and laborious with high error rate.To solve this problem,in this pa-per,a broken rice dataset was created.The dataset consisted of 2 435 images and corresponding label files,contai-ning three categories.In this model,the ShuffleNetv2 lightweight structure was introduced as the feature extraction structure,greatly reducing the number of parameters of the model.On the basis,the BiFPN structure was introduced as the feature fusion structure,and α_IoU was used as the regression box loss to improve the loss function.Experi-ments showed that the accuracy of the improved model reached 98.9％,0.3％higher than that of the original YOLOv5,and the number of parameters and calculation were also reduced by over 85％than that of the original mod-el.Compared with YOLOv3,SSD,RestinaNet and FasterRCNN,the improved model was 0.4％,33.3％,27.9％and 27.2％higher in accuracy,respectively.The relevant datasets are available at https://github.com/THFrag/broken-rice-detection.
NETL
NSTL
万方数据
