Performance Evaluation of a Real Time Video Surveillance System

Abstract

This paper presents a thorough introduction to the real time video surveillance system which has been developed at Bosch Corporate Research considering robustness as the major design goal. A robust surveillance system should especially aim for a low number offalse positives since surveillance guards might get distracted by too many alarms caused by, e.g., moving trees, rain, small camera motion, or varying illumination conditions. Since a missed security related event could cause a serious threat for an installation site, the before mentioned criterion is obviously not suf cient for designing a robust system and thus a low number offalse negatives should simultaneously be achieved. Due to the fact that the false negative rate should ideally be equal to zero, the surveillance system should be able to cope with varying illumination conditions, low contrast and occlusion situations. Besides presenting the building blocks of our video surveillance system, the measures taken to achieve robustness will be illustrated in this paper Since our system is based on algorithms for video motion detection, which has been described e.g. in [1], the previous set of algorithms had to be extended to feature a complete video content analysis system. This transition from simple motion detection to video content analysis is also discussed in the following. In order to measure the performance of our system, quality measures calculated for various PETS sequences will be presented.

Boat Speed Monitoring Using Artificial Vision

“…Different statistical approaches has been proposed such as [4] and [5]. The algorithm described in this article draws on Alsaqre's [6] method to calculate a reference frame as the union between the current frame and the previous reference frame.…”

Performance evaluation for tracking algorithms using object labels

“…The evaluation technique presented in [4] determines the threshold from the distance matrix between the centroid of the bounding box for the ground truths and the result of the tracking algorithm. This threshold is then used to find the correspondence between the result of the tracking algorithm and the ground truth to compute False Positive Track Error, False Negative Track Error, Average Area Error, and Task Incompleteness Factor measures.…”

“…This threshold is then used to find the correspondence between the result of the tracking algorithm and the ground truth to compute False Positive Track Error, False Negative Track Error, Average Area Error, and Task Incompleteness Factor measures. The metrics presented in [4] does not measure the performance of the tracking algorithm in regards to objects labeling and occlusion detection.…”

“…After reviewing the performance evaluation metrics proposed in [3], [4], and [5], we found that these performance evaluation metrics focus on object detection with little reference to how objects are labeled. From subjectively evaluating the performance of tracking algorithms it was also discovered that multiple labels of the same object leads to significant errors in post-tracking-process.…”

“…The selection of [3] over [4] and [5] is due to the fact that its metrics are more comprehensive to the criteria for tracking algorithms. The result of evaluating [1] and [2] for the same six videos using [3] is presented in Table 2.…”