Recognition of blister blight disease in tea leaf using fully convolutional neural networks

Abstract

Tea plantation contributes significantly to the agricultural economy of India. Automatic tea leaf disease detection is beneficial when compared to manual detection which is not only a tedious grueling task but also less accurate and time consuming. This paper presents an alternative image segmentation technique that can be used for automatic detection and classification of blister blight diseased tea leaf using a fully convolutional neural network (CNN) based method to segment blisters in a tea leaf image. The suggested technique proves to be beneficial in monitoring large fields of tea crops.