Smart crossroad surveillance in urban areas

Videos are taking more and more place in the type of data we store on our phones, laptops and other electronics. Technology is advanced to a point we are able to detect, recognise and track vehicles, people or anything that come to our mind. All these algorithms are based on decoded videos. What if it was possible to use the encoded video format and do exactly the same ? Could it perform better ? What about privacy of the surroundings ? The encoded format is quite different than the decoded format, creating a new challenge. The goal of this thesis is to solve the problem, to see if encoded videos are one solution that could lead to more privacy.