Tobacco leaf and tobacco stem identification location detection method based on YOLOV3 network

Aiming at the problem that tobacco leaves are difficult to disperse in the automatic tobacco grading system, a tobacco stem recognition and location model based on YOLOV3 convolution neural network was designed. The algorithm was optimized in network architecture, data processing, feature extraction and other aspects. The robustness of the network was improved by collecting pictures of different grades of tobacco leaves and carrying out data enhancement processing. The tobacco stem parts were labeled by LableImag software, and the network was trained under PyCharm platform. Through comparative analysis, the recognition rate and detection rate of the model were better than SSD detection model. The detection mAP was 90%, and the average detection time was 0.053 s.