Edge computing enabled attendance and access validation system using RFID, facial recognition and line cross detection for payroll integration

Existing attendance and payroll management systems based on Radio Frequency Identification (RFID), Biometrics or cloud infrastructure possess critical vulnerabilities such as proxy marking, card cloning, spoofing and uninterrupted internet dependency. This exposes organizations to fraudulent access as well as data integrity risks. This paper proposes ECHO, a hybrid edge-cloud attendance and access validation system. This system is designed to enforce a secure and multi-layered authentication for payroll integration. It uses a triple-layer verification pipeline consisting of RFID card validation with secrets, a dual biometric face recognition via InsightFace on both wired camera as well as CCTV stream and a physical line cross detection using YOLOv8 and ByteTrack. The data is processed on Raspberry Pi Edge Node with a midnight batch synchronization to cloud storage. Experimental evaluations across multiple trials yielded a False Acceptance Rate of 0%, True Acceptance Rate of 86.67% and overall face recognition accuracy of 93.33% when operating without internet connectivity. The average latency of face recognition is approximately 918 ms. This demonstrates a reliable, cost effective as well as internet resilient solution for attendance and access tracking which can be eventually integrated with a payroll management system.