A recent research at the Binghamtom University aims at analyzing surveillance footage more closely. The team at the New York State University has developed a hybrid lightweight tracking algorithm which makes it easier to follow suspicious activities from camera records.
The algorithm is known as Kerman (Kernelized Kalman filter). For an effective implementation of the experiment the research team used single board computers (SBCs) which monitor human activities. SBCs imbedded with Kerman algorithm proved to be more perceptive to suspicious actions and ensured increased surveillance coverage. After detecting suspicious incidents from live streaming videos, it raises an alert immediately.
Allegedly, the algorithm does not track, identify or record actions of a specific person and thus ascertains high-level of privacy within a secure system. In the future, the algorithm will be rammed into more advanced hardware systems to improve its performance.