Enterprise Asset Tracking Using YOLOv7 for Visual Recognition and RFID Signal Verification
Abstract
Efficient resource management and transparency in the operations of large organizations are essential with the use of enterprise asset tracking. Integration of RFID technology and visual recognition can present an effective solution to identity and verification of assets in real-time. Nonetheless, the current systems are usually victims of the drawbacks like low precision in identifying objects in changing conditions of light and failure to authenticate those assets that do not have a good response frequency using RFID. The consequences of these problems are misidentification, lost inventory, and impaired visibility of assets. To solve these issues, this paper suggests a hybrid system named as "YOLO-RFID TrackNet" that combines the YOLOv7 to provide high accuracy visual recognition and RFID signal verification to provide dual layer asset validation. YOLOv7 is a cutting-edge object detector that is used to detect and categorize assets with high precision using surveillance feeds, and RFID module cross validates the detection with the relevant asset IDs in real time. Experimental findings indicate that YOLO-RFID TrackNet is better at detecting assets and improving verification errors by 18% and 25% respectively than traditional systems. These results indicate that a combination of a high-level deep learning model and RFID checking can become a viable approach to scalable, smart, and correct asset tracking of the enterprise.