Area Ratio-Based Box Counting Algorithm for Intelligent Inventory in Stacked Goods Environments

Inventory management plays a critical role in modern warehouses, but traditional manual counting methods are inefficient and error-prone, especially for large quantities of goods. With advancements in AI, IoT, and computer vision, automated systems have emerged to improve accuracy and efficiency. This paper proposes a novel Area Ratio-Based Box Counting Algorithm (ARBCA) for intelligent inventory in three-dimensional warehouse environments. The algorithm leverages instance segmentation to generate masks for stacked goods and uses area ratios for precise counting. We introduce a customized dataset, the Tobacco Cargo Stacking Dataset (TCSD), and demonstrate the algorithm's effectiveness, achieving an overall inventory accuracy of 93.90% on the TCSD dataset with the Mask2former model. The method addresses challenges such as occlusion and stacking in dark warehouse conditions, showing promise for real-world applications.