Hyperconnect is researching various technolgies to effectively implement data processing through machine learning on mobile devices, servers and browsers. HyperML is a platform based on deep learning-based image processing, and consists of a mobile-optimized inference engine and pipelines for data processing and training.
*Real-time computer vision solutions based on pre-trained models will be available as SDKs on Android / iOS platforms in the near future.
Mobile-optimized deep learning engine
The HyperML inference engine runs on mobile devices. Optimized for the mobile environment, which is typically characterized by limited memory and processing power, the technology ensures real-time results without the data ever being processed by the server.
Artificial intelligence-based image recognition
We have developed a deep learning-based, mobile artificial intelligence system which can, among other things, recognize the current environment wherein the user is situated or filter out inappropriate imagery. This technology is already being used in Azar to provide a more satisfactory matching experience for the users.