Performance evaluation of object detection algorithms for video surveillance

Abstract

In this paper, we propose novel methods to evaluate the performance of object detection algorithms in video sequences. This procedure allows us to highlight characteristics (e.g., region splitting or merging) which are specific of the method being used. The proposed framework compares the output of the algorithm with the ground truth and measures the differences according to objective metrics. In this way it is possible to perform a fair comparison among different methods, evaluating their strengths and weaknesses and allowing the user to perform a reliable choice of the best method for a specific application. We apply this methodology to segmentation algorithms recently proposed and describe their performance. These methods were evaluated in order to assess how well they can detect moving regions in an outdoor scene in fixed-camera situations.

Evaluation of on-line quality estimators for object tracking

“…Consequently, failure of tracking algorithms is inevitable in real tracking systems. Ground-truth information is typically used for the evaluation of a tracking algorithm [1]. However, ground-truth annotations are very expensive to produce and therefore they usually cover only a small portion of video sequences and therefore a small percentage of data variability.…”

A Literature Survey on Animal Detection Methods in Digital Images

“…In [10], threshold segmentation method based on the pixel values is performed. However, in this technique, researchers should carefully choose the threshold value as they also should consider the negative value obtained at certain pixel point by direct subtraction.…”

“…intensity 0). As stated in [10], it is difficult to select the threshold accurately as the background image periodically changes.…”

Track validation using gradient-based normalised cross-correlation

“…A lost track is recorded whenever the overlap between the tracker's rectangle and the object's ground truth bounding box falls below 10% of the area of the latter [13].…”