PROJECT TITLE :

Evaluation of Keypoint Descriptors Applied in the Pedestrian Detection in Low Quality Images

ABSTRACT:

Pedestrian detection is a basic task in video surveillance for systems as of driver assistance systems, tracking pedestrian, detection of anomalous behavior, among others. Native options detectors and descriptors are widely utilized in several computer vision applications and several strategies have been proposed in recent times. Performance evaluation of them is a tradition in pc vision; but, there's a spot comparative of traditional keypoint descriptors like SIFT, SURF and QUICK against recent and novel local feature extractors like ORB, BRISK and FREAK in low quality pictures, as a result of when the number of pixels representing an object is low, the ability to recognize the article is reduced. This article aims to present a scientific and comparative study of the performance these native features detectors and descriptors in pedestrian detection in four real databases, all in an urban environment.