Connected security cameras are almost everywhere. We used them to protect us and our belongings. In a surveillance context, we normally depend on third-party companies to process security images and to generate alarms and warnings. Potential privacy risk exits if those images are evaluated by humans. On the other hand, automatic methods need heavy processing capacities in order to reduce false positive results.
P2Surveil will try to solve those issues by:
-Being fully open source (you will not depend on third-party companies to protect your surveillance data).
-Using a two-layer processing approach to identify threads: first layer will be local camera’s processing unit (image processing layer) and second layer will be a P2P A.I. network (pattern recognition and alarm generating layer). In this way, no image need to be send out of the local context, reducing privacy risks, and increasing computation capacities.