Transforming Higher-Ed Campus Operations with Computer Vision
Manual attendance logging across sprawling multi-department academic universities consumes thousands of faculty hours annually and causes severe transit bottlenecks during class changes. For the Indian Institute of Information Technology (IIIT) Kalyani, Elegensoft designed and deployed an autonomous, touchless facial recognition system capable of processing high-volume student crowds in real time.
Architectural Engineering & Edge Computing
Deploying AI at campus scale required an edge-first architectural approach to circumvent campus bandwidth constraints. Wide-angle overhead IP cameras stream low-latency feeds directly into localized NVIDIA Jetson inference edge nodes.
- Sub-80ms Edge Extraction: Utilizing customized lightweight YOLOv8 models combined with FaceNet vector embeddings.
- Anti-Spoofing Liveness Detection: Multi-frame micro-motion analysis to prevent 2D photo playback spoofing.
- Cross-Platform Mobile App: Built with Flutter for faculty to monitor class rosters and verify real-time check-ins.
- Autonomous LMS Sync: Secure REST webhooks dispatching verified attendance records directly into Moodle LMS.
“The facial AI platform deployed by Elegensoft transformed morning lecture transitions. Over 2,500 students check in effortlessly within seconds.”
Academic Dean, IIIT Kalyani
Key Production Metrics & Outcome
Within the first semester, administrative manual overhead was reduced by 80%, recovering 350+ faculty lecture hours. The model consistently maintains a 99.5% recognition accuracy under varying illumination conditions.