Intelligent Inventory Management in Retail Stores Using Four State of the Art Object Detection Models

One of the most applicable technologies in the context of image processing and machine learning is object detection and tracking. In addition to its use in everyday life, business management development and improvement are two areas where it is particularly useful. Today’s industry demands that inventory management and space allocation be done effectively as there are significantly greater chances of operator error associated with manual tracking. A firm will require additional time to track each item in its inventory as it grows, which increases the likelihood of errors like items may get skipped or wrongfully recorded payments. This paper mainly represents distinct algorithms for automating the inventory management system for retail stores using object detection models. The following object detection models are used in the proposed work—YOLO V7, YOLO V5, FASTER R-CNN (Region-Based Convolutional Neural Network), and MOBILENETSSD. These algorithms are being implemented and compared by mAP, or mean, average, and precision, and from that, the algorithm that works best for tracking is determined. Also, a webapp is created as a medium for visualization of the detected objects. Overall, stock management in smart cities utilizing object detection is a promising use of AI and computer vision that could enhance the effectiveness and sustainability of supply chain management used in retail and wholesale shops in metropolitan regions.