Robust tracking of people in crowds with covariance descriptors

Abstract

In order to control riots in crowds, it is helpful to get the ringleader under control. A great support to achieve this task is the capability to automatically track individual persons in a video sequence taken from a crowd. In this paper we address the robustness of such a tracking function. We start from the results of a previous evaluation of tracking methods, where a so-called Covariance-Tracker was found to be most appropriate. This tracker uses covariance matrices as object descriptors, as proposed by Porikli et al. The set of all covariance matrices describes a Riemannian manifold that is used to compare and update the covariance descriptors during tracking. We propose Covariance-Tracker adaptations to improve its performance. Furthermore, we summarize the performance evaluation results of the original method and compare these with the results of the adapted one. The result is a robust method for tracking people in crowds which can improve situational awareness

Robust person and object tracking in LWIR and VIS based on a new template matching method

“…13,[21][22][23][24] Modern fusion approaches try to combine different image aspects or tracking methods in order to fuse their sets of applications and to build up a more all-purpose object tracker. [25][26][27] But, of course, the overall performance of a fusion method is limited by the performances of the components.…”

“…13,14,22,[28][29][30][31][32][33][34][35][36] Furthermore, some authors use optical flow 13,37,38 or -if the objects are not too small -modern feature sets like SIFT, 39 SURF, ORB, FAST, BRIEF, BRISK and so on. 22,[40][41][42] Other methods use image gradients and moments, 25 similarity measurements based on the joint feature-spatial space 43 or do kernel-based tracking, 44 for example.…”

Person re-identification across aerial and ground-based cameras by deep feature fusion

“…Covariance descriptors were originally introduced for detection and classification tasks in [3]. They were then successfully applied to tracking [7] and have been shown to be well suited for person tracking in aerial images [8,9]. They also yield high accuracy when used for appearance-based person re-identification [10,11].…”

Video-tracker trajectory analysis: who meets whom, when and where

“…are handed over from one person to another in crowded environments like railway stations, airports or busy streets and places etc.. The input to the trajectory analysis comes from a multi-object video-based tracking system developed at IOSB which is able to track multiple individuals within a crowd in real-time [1]. From this we calculate the inter-distances between all persons on a frame-to-frame basis.…”