The problem
Smart-city camera analytics usually stream video to a server, which costs bandwidth and raises privacy concerns. Processing at the edge and sending only counts solves both, but small devices must keep up with the frame rate.
Who needs it
Parking operators, campuses, municipalities, shopping centres.
Idea of the approach
- Public parking datasets (CNRPark-EXT, PKLot) plus a few recordings on campus (with permission, no faces kept).
- Occupancy classification per spot and vehicle counting with a light detector.
- Edge pipeline on a Jetson/Raspberry Pi that only publishes numbers (MQTT) to a live dashboard.
- Measure accuracy, frames per second, bandwidth saved.
What you will build and deliver
- Working edge pipeline and dashboard.
- Evaluation report and paper.
Skills needed
Python, OpenCV, basic web dashboard.
Research contribution & where to publish
Contribution: accuracy/latency/privacy evaluation of edge-only video analytics. Venues: IEEE SMARTCOMP, IEEE ISC2, Sensors, IEEE Access.
Key references — start reading here
- G. Ananthanarayanan, P. Bahl, P. Bodík, K. Chintalapudi, M. Philipose, L. Ravindranath, S. Sinha. Real-Time Video Analytics: The Killer App for Edge Computing. IEEE Computer, 2017.
- G. Amato, F. Carrara, F. Falchi, C. Gennaro, C. Meghini, C. Vairo. Deep learning for decentralized parking lot occupancy detection. Expert Systems with Applications, 2017.
- M. Satyanarayanan. The Emergence of Edge Computing. IEEE Computer, 2017.
Interested?
Send the topic title and the courses you have taken.
Contact Dr. Hamza Djigal