Enhanced Coffee Bean Defect Detection Using Super-Resolution GANs and Deep Learning

Image Super-Resolution (SR) is a key challenge in computer vision, focusing on enhancing low-resolution images while preserving details. This paper presents a novel method using Generative Adversarial Networks (GANs) for improving image quality, specifically targeting coffee bean defect detection. The proposed SR-GAN model includes a generator that upscales low-resolution inputs and a discriminator that differentiates between generated high-resolution images and real ones. Trained on a coffee bean dataset, the model optimizes loss functions to balance perceptual quality and fidelity, capturing subtle defects like cracks and discolorations. Enhanced images are processed by two Convolutional Neural Networks (CNNs): one identifies broad defect categories, such as defective black beans, brown beans, broken beans and non-defective high-quality beans. This layered approach ensures more precise defect identification and classification, overcoming limitations of previous methods that struggled with subtle defects. Performance evaluations show that SR-GAN and CNN significantly outperforms existing models in resolution enhancement and classification accuracy, offering a robust solution for improving quality control in coffee production.