AGRI-Connect: An Ai-Driven Unet–Vision Transformer Framework For Disease-Aware Crop Quality Grading And Direct Agricultural Marketplace Integration

Small and marginal farmers in developing economies face challenges such as delayed crop disease detection, subjective quality assessment, and non-transparent pricing due to intermediary-dominated markets. To address these issues, this paper presents Agri-Connect, an AI-driven digital platform that integrates automated crop disease detection, quality grading, and a direct farmer–consumer marketplace. The proposed system employs a hybrid deep learning architecture, combining UNet-based semantic segmentation for precise diseased region extraction and a Vision Transformer (ViT) for robust disease classification and severity analysis. Experimental evaluation was conducted on a combined dataset consisting of ICAR images, drone-captured imagery, and real- world field images under varying environmental conditions. The proposed framework achieved a disease classification accuracy of approximately 94% and reliable quality grading performance across multiple produce categories, outperforming conventional CNN-based approaches. A multilingual, voice-enabled interface and AI- powered chatbot enhance usability for low-literacy users, while an integrated real-time marketplace enables transparent, quality-based pricing and direct trade. Agri-Connect demonstrates the practical potential of linking AI-verified crop analysis with digital market access to improve farmer income, transparency, and sustainable agricultural practices.