A Gan-Based Approach for Automated Plant Disease Classification Using Image Processing Techniques

Plant diseases represent a significant danger to agricultural output, frequently resulting in considerable crop losses. Timely and precise detection of such diseases is essential to guarantee the production of safe and nutritious food. Due to the insufficient skill of numerous farmers in identifying diseases-particularly when symptoms are vague or difficult to discern-automated plant disease detection utilising artificial intelligence has emerged as a crucial instrument. This research presents an AI-based framework for the more effective diagnosis of plant diseases. The procedure commences with image preprocessing, wherein the colour space is converted from RGB to HSV, succeeded by image segmentation employing Kapur's thresholding method to delineate the affected leaf areas. Colour features are obtained via Colour Histograms, whilst shape features are generated using statistical descriptors. A deep learning methodology utilising GAN is implemented, wherein two neural networks are concurrently trained and evaluated against a conventional CNN. The proposed method attained a peak accuracy of 97.34 %, surpassing current deep learning models. These findings highlight the promise of incorporating GANs and image processing methods into automated plant disease detection systems to provide farmers with prompt and accurate diagnosis.