Creative Commons source reading · Data quality and algorithm evaluation
Deep convolutional neural network-based automated identification and classification of mungbean foliar diseases
Shail Bala; S. I. Harlapur; Aditya Kamalakar Kanade; M. P. Potdar; Gurupada B. Balol; V. B. Kuligod; Lingareddy Usha Rani
Originally published in Front Artif Intell · 2026-09-03
From the article
Selected source text. JournalOS removes markup and shortens excerpts where indicated. The full article provides the complete context.
Abstract
Early and accurate detection of plant diseases is critical in precision agriculture to improve crop management and yield. Mungbean (Vigna radiata L.) is highly susceptible to several foliar diseases, including yellow mosaic, powdery mildew, leaf crinkle, and cercospora leaf spot, which cause substantial productivity losses. Despite expanding applications of deep learning in plant disease diagnosis, systematic multi-architecture evaluation for mungbean disease classification under natural field conditions remains limited. This study addresses this gap by evaluating five state-of-the-art deep convolutional neural network (DCNN) architectures on a large-scale, field-acquired mungbean dataset that captures real-world variability across environmental conditions and disease severity levels, distinguishing it from controlled laboratory…
Discussion
This study provides a statistically robust, multi-perspective benchmarking framework for the automated identification of major foliar diseases in mungbean using in-field leaf imagery. Unlike traditional deep learning frameworks in agricultural computer vision that rely solely on empirical point-estimates of accuracy, this workflow integrates pairwise hypothesis testing via McNemar's test with Holm-Bonferroni corrections alongside extended metrics such as Cohen's kappa and macro-standardized F1-scores. This mathematical framework provides stronger evidence that the observed performance differences are unlikely to be due to random sampling variation in feature extraction capability rather than random sampling artifacts, addressing a critical gap noted in recent crop phenotyping literature (Richter et al., 2025). Among the evaluated…
Conclusion
This study presented a comprehensive deep learning framework for automated classification of major foliar diseases in mungbean using field-acquired image data. Five state-of-the-art DCNN architectures, namely VGG16, VGG19, ResNet50V2, DenseNet121, and InceptionV3, were systematically evaluated under identical transfer learning conditions. The proposed framework incorporated rigorous dataset preparation, post-split data augmentation, class-weighted learning, and extensive performance evaluation using multiple statistical and machine learning metrics. The results demonstrated that all evaluated architectures were capable of accurately discriminating among healthy leaves, yellow mosaic disease, powdery mildew, leaf crinkle, and cercospora leaf spot under natural field conditions. The major contributions of this work include the…
Editorial reading guide
- What data created and tested the model, and was evaluation independent?
- What is the cost of a false positive, false negative or out-of-distribution decision?
