Agriculture remains a primary source of livelihood for a large share of the Indian population and is essential to food security and national economic development. Variations in climate, soil condition and local environment increase the risk of crop disease, and because the earliest symptoms of most infections appear on the leaf, an infection that is not identified in time spreads through the plant and across the field. Manual inspection by trained experts is slow, costly, inconsistent and impractical over large cultivation areas, so automated detection has become a necessity rather than a convenience. This paper reviews the development of automated plant leaf disease detection from classical image processing and machine learning through deep convolutional architectures to the vision transformer, hybrid and deployment-oriented models reported up to 2026. Thirty-six representative studies are organised into a four-family taxonomy and compared in terms of technique, dataset, crop coverage and reported accuracy. A generic five-stage detection pipeline — image acquisition, pre-processing and augmentation, segmentation, feature extraction and selection, and classification — is presented together with a model architecture in which convolutional, transfer-learning and transformer backbones are interchangeable. The review shows that reported accuracy on the standard benchmark has risen from the high eighties for early convolutional and image-processing systems to above ninety-nine per cent for recent transformer and lightweight convolutional designs, but that this progress is measured almost entirely under laboratory conditions. Persistent gaps are identified in field-condition robustness, dataset diversity, class imbalance, earlystage and multi-disease detection, computational cost at the edge, and interpretability, and these gaps are consolidated into a set of directions for future work.

Atul Raj Dhairya, Prof. Nagendra Patel,"A Comparative Study of Deep Learning Architectures for Plant Leaf Disease Detection: A Systematic Review" International Journal of Advanced Multidisciplinary Research and Educational Development, V2(5): Page(392-398) September - October 2026. ISSN: 3107-6513. www.ijamred.com. Published by Scientific and Academic Research Publishing.