Diagnosis of leaf health using grape leaf thermal imaging and convolutional neural networks

Crop health has been an important aspect of farming as the produce and income depend on it. Large amount of research work has been carried out in crop health diagnosis. This paper suggests the using thermal imaging for identifying the occurrence of disease on the crop by using convolutional neural networks approach. Thermal images are initially preprocessed using Thresholding and morphological operations. Furthermore, group convolution strategy is used to train the model with the use of a number of kernels for processing the data.