Plant Leaf Disease Identification Using VGG, ResNet, Inception-ResNet, and DenseNet Models with Optimized Image Enhancement

Plant diseases cause a significant threat to global agriculture, reducing crop yield and quality. This research presents a deep learning-based image analysis approach for the accurate and early detection of plant leaf diseases. The methodology integrates advanced image preprocessing techniques, including Lab color space transformation, background correction, blurring, and mask generation to improve image quality and isolate diseased leaf regions. Five pretrained convolutional neural networks-VGG16, VGG19, ResNet50, Inception ResNet V2, and DenseNet121-were fine-tuned using TensorFlow and Keras frameworks. Each model was evaluated based on training and validation accuracy and loss. Among the models, ResNet50 achieved the highest validation accuracy of 95.92 %, demonstrating greater feature extraction and classification performance on the Plant Village dataset. The results confirm that combining sophisticated image preprocessing with powerful pretrained CNN architectures significantly boosts disease classification performance. This technique enables early- stage monitoring and diagnostic analysis of plant diseases, facilitating timely intervention and potentially reducing agricultural losses.