The shape characteristics of the main veins of tobacco leaves are one of the reference standards for tobacco grading.The article improves on the U-net model and applies it to the extraction of main veins on the surface of tobacco leaves.The model sets side outputs in all upsampling layers and constrains these side outputs to improve the recognition performance of the model.The U-shaped segmentation network was pruned while ensuring model performance,reducing model redundancy and improving model inference speed.The comparative analysis of the experimental results verified the effectiveness of the model in the task of segmenting the main veins on the surface of tobacco leaves.
关键词
烟叶主脉/语义分割/U-net
Key words
main vein of tobacco leaves/semantic segmentation/U-net
维普
万方数据