An Approach for Rice Bacterial Leaf Streak Disease Segmentation and Disease Severity Estimation

Rice bacterial leaf streak (BLS) is a serious disease in rice leaves and can seriously affect the quality quantity of growth. Automatic estimation severity crucial requirement agricultural production. To address this, new method (termed BLSNet) was proposed for BLS lesion recognition segmentation based on UNet network semantic segmentation. An attention mechanism multi-scale extraction integration were used BLSNet to improve accuracy We compared performance with that DeepLabv3+ as benchmark models It found model demonstrated higher class accuracy. A preliminary investigation carried out our results, it has strong potential be reliable automatic estimator severity.