As tomatoes are one of the most frequently consumed crops worldwide, the earlier detection of tomato disease is very important, which is crucial for preventing yield loss and ensuring sustainable farming. Several methods have been developed to detect and mitigate such diseases, helping to protect tomato plants effectively. This project presents a deep learning-based approach that utilizes Generative Adversarial Networks (GANs) alongside autoencoders to detect diseases of the tomato plants from leaf samples. The autoencoder is trained to recognize and reconstruct healthy leaf images, capturing essential features. At the same time, the GAN generates synthetic images of diseased leaves, expanding the dataset to improve model training. Incorporating synthetic data improves adaptability of the model to different leaf conditions, increasing its accuracy. A discriminator is employed to assess the generated images, enabling the model to extract robust features required for precise disease detection. This approach facilitates real-time identification of plant diseases, reducing dependence on manual inspections. Recognizing infections in the initial stages helps minimize crop losses and supports sustainable agriculture. By integrating GANs with autoencoders, this method offers an effective solution to one of agriculture’s key challenges: timely recognition of diseases in crops.