Sensing and Spectroscopy Lab

Deep learning methods and image processing techniques were applied to automatically identify 3 diseases and 4 pests on tea leaf images collected under practical field conditions.

The proposed approach applied faster region-based convolutional neural network (Faster R-CNN) to localize the diseased regions and identify the types of diseases/pests.

Model performance achieved a precision of 77.5%, a recall of 70.6%, and an accuracy of 89.4%.