Paper Title
A Design Overview on Multi Person Tracking using Deep Learning

Abstract
The mechanism of locating multiple objects over a sequence of frames (videos) is referred as multi object tracking. Tracking by detection is best method of choice for approaching noisy and imprecise measurements, long term occlusion and target interactions as tracking by detection is able to reduce complexity dynamically by taking only restricted set of measurements. Tracking by detection method avoids some difficulties like drift, since it is self initializing and is often able to recover occlusion for extended period of time. Large number of dynamically moving persons is detected by the algorithm in complex scenes with occlusions, only makes use of information from the past. Keywords - Tracking by Detection, Convolution Neural Network, Online and Offline Multi Person Tracking (MTP).