Title: Analytical method for plant disease detection in leaf using EfficientNet architecture

Authors: Ashish Gupta; Deepak Gupta; Jaimala Jha; Shubangi Vairagar; R. Vadivel; Kunda Suresh Babu; Sudipta Banerjee; K.B. Yuvaraj

Addresses: Department of Information Technology, Institute of Technology and Management, Gwalior, Madhya Pradesh, 474001, India ' Department of Computer Science and Engineering, Institute of Technology and Management, Gwalior, Madhya Pradesh, 474001, India ' Department of Computer Science and Engineering, Madhav Institute of Technology and Science, Gwalior Madhya Pradesh, 474001, India ' Department of AI & DS, Dr. D.Y. Patil Institute of Technology Pimpri, Pune, 411002, India ' Department of Artificial Intelligence and Data Science, Nitte Meenakshi Institute of Technology (NMIT), Nitte (Deemed to be University), Yelahanka, Bangalore, Karnataka – 560064, India ' Department of Computer Science and Engineering, Narasaraopeta Engineering College (Autonomous), Narasaraopet, Palnadu(dt) – 522601, Andhra Pradesh, India ' Computer Science and Engineering, Symbiosis Institute of Technology, Pune Campus, Symbiosis International(Deemed University) (SIU), Pune, Maharashtra, 412115, India ' Department of Mechanical Engineering Mangalore Institute of Technology and Engineering, Moodabidre, Mangalore, South Canara, Karnataka, 57500, India

Abstract: This research implements EfficientNet to identify blight diseases in tomato and potato leaves, benefiting farmers in rural India. These crops are vital to both local and global food supplies but are susceptible to diseases like late and early blight, which harm production and farmers' income. The research uses machine learning (ML) and deep learning (DL), leveraging the Plant Village Dataset with 3981 training images and 2255 test images. After hyperparameter optimisation, the model achieves an accuracy of 99.36%, precision of 99.71%, sensitivity of 99.45%, and an F1-score of 99.43%. The EfficientNet model identifies key features such as leaf shape, infection areas, and green regions, improving understanding of its predictions. These findings could enhance convolutional neural network (CNN) algorithms for crop disease classification, boosting agricultural practices and crop yields in India.

Keywords: deep learning; EfficientNet; SHAP; SHapley Additive explanations; blight disease; potato; tomato.

DOI: 10.1504/IJDATS.2025.150941

International Journal of Data Analysis Techniques and Strategies, 2025 Vol.17 No.4, pp.345 - 370

Received: 08 May 2024
Accepted: 19 Sep 2024
Published online: 05 Jan 2026 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article