ML optimised for running on a for low-end device:
There is nothing like coming home after a long day. Our home is our sanctuary, the place where we can feel at peace. Protecting your home is not longer only the job of the police or a good dog. In the new social era we are living, we need smarter solutions, smarter assistants, that can recognise and identify easily our friends, from intruders or unwanted presence. Such solutions should adapt to the home specifics, such as homes with noisy environments, random schedules, large variety of sounds, variety in the types of visual visitors - and use complementary cues from the home context or person’s preferences, for yielding better predictions. Flare is a unique approach to home security. It protects user's home all by itself, recognises and identifies friends from foes and is able to take decisions automatically - sending the police, or considering it as a false alarm. Using embedded machine learning algorithms powered by hardware sensors, Flare can understand what is happening in the world around it, and predict dangerous events occurring at home. We are analysing both audio (dog barking, glass breaking, steps, speech, alarm, etc.) and video (humans, dogs, cats, faces, fire, etc.) in a parallel manner, to classify events, and learn patterns from them in time. All this intelligence working on a small embedded device was highly optimised to work in a collaborative way, so the result of one AI algorithm would help the other algorithms generate better predictions. In a parallel to the mobile industry, we were solving challenges that mobile manufacturers have today running AI on small devices (e.g IphoneX’s face recognition, voice recognition (Siri, Google now)). Top technical limitations and challenges, together with various solutions, usability concerns in IoT, state-of-the-art of the industry will be presented and discussed.
George is a full-stack engineer, with a great passion for new technologies, hardware and machine learning. During the years, he was working in different positions, such as web/firmware developer, mobile developer, team leader, branch manager and lately entrepreneur. He enjoys getting the best learnings from all the positions he has been through and the people he met. Currently, George is CTO of BuddyGuard, a home security product that disrupts the smart home market through special features powered by artificial intelligence (face recognition, voice analysis, pet detection, etc.). At BuddyGuard, he is managing the technical side, which involves many challenges on data security, audio and video machine learning algorithms, distributed cloud systems scalability, mobile phones integrations and hardware electronics integrations.