Computer Vision News - November 2016

Every month, Computer Vision News reviews a successful project. Our main purpose is to show how diverse image processing applications can be and how the different techniques can help to solve technical challenges and physical difficulties. This month we discuss a fully automated shareable surveillance system . Do you have a project in computer vision and image processing? Contact our consultants . Last month we have reviewed recent breakthroughs brought about by computer vision in the surveillance area and how RSIP Vision’s algorithms contributed to this major innovation. This month we are going to see what breakthroughs can we expect in the next few years and what is needed to make that happen. Building on the innovative impulse given by big data and deep learning to track events which previously could not be brought to our attention, the upcoming breakthroughs will be given by cloud processing . This will happen in two ways: on one hand, with powerful computers, such as are not yet available, in specific location; on the other hand, with information which is shareable between different points (cameras, users and the like). Scalable processing , starting at the level of streets, can reach sizes of the order of cities and nations. All information which sensors can detect in the street with RGB-IR cameras, sound, radar, aerial imaging, proximity switches and all Internet of Things items can be aggregated into data. This data is then processed and common patterns or known behaviors are automatically detected, recognized and filtered (e.g. a truck delivering on fixed days or human faces passing by every day). The result is that any irregular, significant pattern (like a suspect individual, even in cases of terrorist actions ) is collected and shared with other cities and countries as needed, without any human intervention . Parts of this system are already developed, while the main challenge to its full realization comes from the technicalities of installing the sensors. Data collection points should be installed and information should be streamed through large bandwidth communication systems, connected to powerful servers on the cloud and responsible for running all the deep learning algorithms in real time. RSIP Vision knows how to handle big data, develop deep learning and other smart algorithms to process this data and how to backup the system with high-end rule-based heuristic algorithms at any level of this scheme. Project Fully Automated Shareable Surveillance Systems Computer Vision News Project 59