AI and computer vision platform for facial recognition
The service provides the possibility to recognize faces using video streams from HTTP/RTSP connections with outdoor cameras. The system can identify persons, their age and gender. It easily integrates with existing camera systems. Using the AI and Computer Vision technology it can gather and analyze massive amounts of data.
The client had an old semi-working prototype system running on Django templates with no front-end frameworks. The task was to create a new visual template, run the code, add the ability to view all cameras connected to the system in real time, then rewrite the system using the framework.
- Recognition dashboard
- Video 2-Factor Authorization (V2FA)
- iOS Check-in App
-
We delivered recognition dashboard MVP in 2 months
-
We delivered V2FA MVP in 1 month
-
We delivered iOS Chek-in App MVP in 2 weeks
We used Django REST API as the basis for the backend of the new version of the system, Angular framework to build the frontend application. This combination allows rational use of hardware capacity. It also optimizes system performance, allowing easy scaling in the future.
Recognition Dashboard:
- aiortc, Python and OpenCV for implementing the app and decoding cameras stream video,
- Amazon Rekognition API for facial recognition
- amCharts for building visual analytic report of the collected data
For V2FA were used:
- OpenCV for recognizing the presence of a face in a frame
- OAuth for integration with the external sites
- Tensorflow for gesture recognition when logging
In iOS Check-in App we used:
- Swift for building iOS app architecture
- Apple Vision for recognizing the presence of a face in a frame
- Zoom SDK to enable calls to technical support
