Research on Warehouse Inventory Counting Application Based on YOLOv8

In industrial enterprises, warehouse management plays a pivotal role in both construction management and production management. Traditional methods for recording inbound and outbound materials often rely on manual operations and barcode scanning, leading to difficulties in counting, time consumption, and omissions. Therefore, we developed an automatic recognition system based on YOLOv8, the system leverages edge computing capabilities and is deployed on lightweight hardware platforms to perform real-time image acquisition of storage shelves or warehouse entry/exit points via cameras. Utilizing the YOLOv8 object detection model, it enables intelligent classification and quantitative analysis of stored items. Additionally, the system incorporates differential comparison and anomaly warning mechanisms, allowing for timely identification of inventory discrepancies. Experimental results demonstrate that the system excels in recognition accuracy, processing efficiency, and deployment flexibility, offering a robust solution for the digital transformation of traditional warehouse management. It also provides strong technical support for enterprises advancing toward intelligent and information-driven operations.