Optimized Surveillance
- 0 Collaborators
This project aims at improving efficient usage of the storage device by keeping relevant data from surveillance. In many scenario a surveillance camera looks at an empty corridor. We aim at recording the feed only when a human is detected in the feed. ...learn more
Project status: Concept
Networking, Internet of Things, Artificial Intelligence
Intel Technologies
Movidius NCS,
Intel Python
Overview / Usage
The surveillance camera usually looks at an empty corridor and write the feed on the disk. This is waste of storage device and there is no log of important data maintained. The surveillance camera is in place to usually keep a check on theft. But if there is no human in the frame it's pointless to record the feed. Thus we came up with a solution architecture where human detection is performed at every frame. This will in turn record the feed with human presence. Other stats can be drawn as well like the number of humans, etc.
Methodology / Approach
Since it's in development phase the following methods were used:
- Looked through various hardware alternatives where human detection can be performed.
- Look through various algorithms to perform the same.
- Test on live feed.
- Deciding on server alternatives and overall cost.
Technologies Used
- TensorFlow
- Keras
- OpenCV
- Raspberry
- OpenVINO