Improved Commodity Supply Chain Performance Through AI and Computer Vision Techniques

Abstract

In the realm of supply chain management, the impact of Artificial Intelligence (AI) tools on optimizing commodity distribution is undeniable. This study presents the transformative potential of AI and computer vision in the field of commodity supply chain management. The capability of AI to reduce yield loss and enhance supply chain efficiency is a growing trend and vision-based commodity defect monitoring can be useful in this regard. We explored the employment of real-time computer vision techniques in supply chain flaw management, which include Detection Transformer (DETR), a type of Vision Transformer (ViT), and compared its performance with the You Only Look Once (YOLO) and other AI models. Computational feasibility is assessed, encompassing various computer vision and AI models, by using a dataset comprising images of commodity items used to substantiate our findings. The obtained results have shown the improved performance of DETR with a detection and classification accuracy of 96%, directly correlating with improved supply chain management. On the other hand, the higher computational burden imposed by DETR makes it less feasible for the higher constrained embedded applications. The practicality of AI algorithms for real-time defect identification reveals promising prospects for integration into supply chain systems. This research underscores AI’s potential to revolutionize commodity supply chain management, extending its benefits to various commodity distribution networks.

Keywords

• Supply chain management
• computer vision
• artificial intelligence
• smart agriculture