ИСТИНА

The paper considers a problem of multiple person tracking. We present the approach to automatic people tracking on surveillance videos recorded by static cameras. Proposed algorithm is an extension of [1] and is based on tracking-by-detection of people heads. It performs data association using Markov chain Monte Carlo (MCMC). Tracklet postprocessing and accurate results interpolation were shown to reduce number of false positives. We use position deviations of tracklets and revised entry/exit points factor to separate pedestrians from false positives. Finally, the paper presents a new method to estimate body position. Our evaluation shows results competitive to modern tracking methods.