Innovative Research Publication International Journal of Innovative Research in Computer Science and Technology 2347-5552 Volume 14 Issue 5 English Sep-Oct - February 2026 N N Y 2026 10 09 Computer Science Genetic Algorithm-Optimized Vision Transformer for Automated Tomato Leaf Disease Classification English Y 40 47 Gunjan Mishra English Y Harsh Dev English N Sudheer Kumar Singh English N https://doi.org/10.55524/ijircst.2026.14.5.5 Automated detection of diseases in tomato leaves will assist in the timely evaluation of the crop health. However, the performance of Vision Transformer (ViT) is highly dependent on the hyperparameters. The presented research concerns a Vision Transformer optimized by a Genetic Algorithm (GA-ViT) for the binary classification of Early Blight and healthy tomato foliage using 2,591 images divided according to a stratified 70:15:15 training-validation-testing framework. The images were resized to 128 x 128 pixels and divided into 256 individual non-overlapping 8 x 8 pieces. A Genetic Algorithm was used to tune the learning rate, dropout rate, feed-forward dimension, number of attention heads and number of transformer blocks with validation accuracy used as the fitness function. For each seed, GA optimization and final model training were performed independently. The test accuracies were 97.94%, 95.63%, and 94.86% for seeds 42, 52, and 62, respectively. The mean test accuracy was 96.14%, with a standard deviation of 1.61 percentage points. The mean weighted precision was 96.18%, weighted recall was 96.14%, and weighted F1-score was 96.13%. The mean test loss was 0.1276. The results indicate the potential of evolutionary hyperparameter optimization to improve ViT-based plant disease classification. English Deep Learning, Genetic Algorithm, Image Classification, Vision Transformer, Tomato Leaf Disease Classification https://ijircst.org/view_abstract.php?title=Genetic Algorithm-Optimized Vision Transformer for Automated Tomato Leaf Disease Classification&year=2026&vol=14&primary=QVJULTE0ODA=