frAPI is a system to automate the detection and identification of faces, with the possibility to generate and manage events and alarms. Also, the software allows the monitoring of streaming in real time. frAPI can be used as an API (Application Programming Interface) locally and/or cloud through HTTP calls.

You may deploy it in one of two ways: on-premise or on the cloud. When using the on-premise version, with a single installer you have the system up and running in your own servers. The cloud deploy may be accessed directly using a URL, username and password. Meerkat also provides a user manual, with general information, and an API documentation via API Blueprint. Our engineering team helps installing and integrating the system into your software.

Yes, for our development version you can access our demo. You'll also find complete documentation. Support happens via email and remote access.

In order for the system to be able to process videos in real time, we recomend an Intel i5 or i7 processor, with AVX instructions and 4GB RAM. GPU is not required. Running the system on this hardware on top of the Linux Operating System you get around 30 FPS per camera, up to 4 cameras. You may increase the number of video streams, but the processing power decreases nearly linearly. For example, 8 camera streams will be processed at 15 FPS; 12 camera streams at 7 FPS; and so on. frAPI scales so that it's able to use all the processing power of servers. For maximum performance, Meerkat recommends using a dedicated server. If you have a different setup, Meerkat's enginnering team will help you with required adjustiments.

No. The system was developed aiming at extracting the best performance of low-end hardware. It's possible to run in GPU for larger scale projects. Meerkat's system is ready for that and will have better performance.

Our system is robust to image variations. However, there are settings that optimize recognitions according to the scenario:
- Open spaces (shopping mall, stadiums): Since there is a lot of exposure to external factors, we advise using high resolution cameras, with focus optimization. Using these cameras you'll also be able to identify people at a greater distance and in agglomerates.
- Closed spaces (schools, universities, office spaces): These are more controlled environments, allowing for the use of lower cost cameras, having at least 720p.

frAPI has 95% accuracy on the Megaface dataset, with a million enrolled faces and response times of less than a second.

frAPI doesn't detect emotions, gender or age. However, these features can be developed on demand by our engineering team. Please reach out to us if you need additional features.