Groundnuts Leaf Disease Recognition Using Neural Network with Progressive Resizing

Main Article Content

T. Kosalairaman, M. Sathyanarayanan, V. Kumaresan, K. Sivasankari


Groundnut is an important oilseed crop worldwide and India is the second largest producer of groundnut. This crop is susceptible to many diseases, which is one of the main reasons for the decline in productivity and quality. Both ultimately lead to a downturn in the agricultural economy. Therefore, a better and more reliable automatic solution for detecting groundnut leaf diseases is needed. In this paper, we propose a deep learning-based progressive resizing model for peanut leaf disease detection and classification tasks. There are five main categories of groundnut leaf diseases: leaf spot, armyworm, wilt, yellow leaf, and healthy leaf. The proposed model was trained with and without stepwise resizing and simultaneously validated using cross-entropy loss.The first dataset of its kind used for training and validation purposes was manually created in the Saurashtra region of Gujarat, India. The resulting dataset was unbalanced because each category had a different number of samples. An advanced focus loss feature was used to solve the dataset imbalance problem. To evaluate the performance of the proposed model, various performance metrics such as accuracy, sensitivity, F1 score, and accuracy were applied. The proposed model achieved an accuracy of 96.bythe state-of-the-art standard12.The progressive resizing model performed better than the traditional cross-entropy loss-based core neural network model.

Article Details