In India, agriculture is pivotal to the economy, and crop diseases significantly impact productivity. While some diseases can be visually identified, many require diagnostic processes, posing challenges for farmers with limited expertise. Modern technologies are essential for providing accurate and efficient solutions to reduce crop losses and ensure food security. This study aims to develop an automated system for identifying and classifying diseases affecting okra leaves using advanced neural network methodologies. The proposed framework utilizes a comprehensive dataset of high-resolution images of healthy and diseased okra leaves. The methodology comprises three primary steps: vein segmentation, graph construction from vein images, and graph classification using a Graph Convolution Network (GCN). Vein segmentation employs a method based on existing literature to ensure precise extraction of vein structures, which are crucial for accurate disease identification. These vein structures are then used to construct graphs, representing the complex patterns of leaf venation. The graphs are classified using the GCN, a powerful neural network architecture known for its efficiency in handling graph-structured data. The GCN’s ability to focus on essential data relevant to the problem significantly enhances its performance. Additionally, the model’s high classification accuracy and compact size make it suitable for deployment in resource-constrained environments. Furthermore, this study conducts a comparative analysis with transfer learning architectures such as VGG16, Inception, and ResNet50. The results demonstrate the GCN model’s superior efficacy in capturing the characteristics of okra leaf diseases, outperforming traditional machine learning methods and advanced CNN architectures. This study underscores the importance of integrating modern machine learning techniques in agriculture, paving the way for future research and advancements in this critical field.