Deep Learning-Enabled Inventory Detection for Facility Management System

Abstract

Facility management benefits from efficient inventory tracking to optimize resource allocation and control costs. In this study, a deep learning-based method is presented for detecting and quantifying office inventory from office images. The proposed system employs YOLOv9 for object detection; however, since YOLOv9 does not include predefined categories for office supplies, a custom dataset is developed. A total of 10,000 images of common office inventory items are collected through web crawling, and 2,000 images are manually annotated using Roboflow to support model training. The dataset consists of 14 office inventory classes, ensuring comprehensive coverage of essential items. Experimental results demonstrate that the model effectively detects and counts office inventory items, providing a reliable solution for automated inventory management. The proposed approach minimizes manual effort, improves tracking accuracy, and enhances the efficiency of facility management operations.