Faceter Fog: Self-hosted power, live recognition, and more
Dear community members, as we promised, the Faceter team is honored to present our product achievements and development results of 2021 alongside midterms plans for the next few months for further 2022 development.
Watch Faceter Fog CEO Update on YouTube clicking this link 👈🏻
1) Live recognition of RTSP streams.
Anyone who is registered in the Faceter Fog camera owner account can add a stream themselves using RTSP real-time streaming protocol applicable to any IP camera. In a real-time stream, you will be able to see detected faces and persons which were created during this recognition process.
This technology allowed us to reduce delay just to 2–3 seconds. Before we streamed via archive:
📍 A camera was streaming to the archive.
📍 Miner’s node was processing the video files from the archive
📍 It took at least 20 seconds before we could understand what faces were recognized.
Now you can use Faceter live recognition algorithms anywhere where real-time face detection is necessary, e.g., at the security gate.
2) Improved recognition accuracy is now available for Faceter Fog users.
We reworked our neural network algorithms to avoid face duplicates and false positive and negative recognition results. We were working to introduce an updated neural network algorithm and move from the Caffe framework we used to TensorFlow, an industry gold standard framework for computer vision and AI. This increased recognition accuracy allows us to count the exact number of unique visitors while before, a single person could have many profiles not linked between themselves.
3) Search by photo.
Now you can search in the list of recognized profiles with a person’s photo. If you upload an image, the system will compare it with all the profiles stored in the list of visitors and match them.
4) Analytics in your camera owner account.
Analytics segregates information of registered visitors by gender/age and displays the total number of visits, unique visits, and new visits recorded. A camera owner can view analytics in his account for all video sources.
5) Try facial recognition by video or image for free.
We updated our landing page and now the algorithms can be tested without registration. Everyone who has a smartphone can shoot a short video with faces in it and then upload it to our landing page to see the data extracted from this video. This technology will also be available through our API.
Now we have great live tech; our goal was to stabilize it and get good recognition results. Within the next few months, we are setting for ourselves a new development challenge:
- UX and UI improvements for clients usability;
We will be happy to hear of any improvement ideas you would like to contribute to us. Feel free to test Faceter Fog and tell us how you think it could be better.
Another development challenge we face is to develop actual recognition parameters for a neural network via numerous tests with different configurations to find the best option suitable for real-world use cases installation.
We are eagerly waiting for your feedback whether it is positive or not! It is all about our steady improvement and mutual collaboration on these ambitious project development goals!
Stay tuned for more exciting updates coming soon!
If you still have any questions, please send us an email to firstname.lastname@example.org or ask our 24/7 support team in Telegram.
Know the people around you,