GPU detektor objektů ve videu

English title

GPU object detector in video stream

Type

software

License

In order to use the result by another entity, it is always necessary to acquire a license

License Fee

The licensor does not require a license fee for the result in some cases

Authors

Kula Michal, Ing., Ph.D. (DCGM)
Zemčík Pavel, prof. Dr. Ing., dr. h. c. (DCGM)

Keywords

object detection, WaldBoost, GPU, OpenCL, LBP, LRD, LRP, CSLBP

Description

Real-time object detection is a core part of various complex algorithms in surveillance systems, traffic monitoring, autonomous vehicles and many other fields. Object detection is one of most computationally demanded part of these algorithms. Especially real-time processing of high resolution, high refresh rate or multiple image-based streams without rescaling or frame dropping are hard to achieve. Software "GPU object detector in video stream" is suitable to achieve these goals thanks to parallel GPU implementation of detection algorithm which includes several optimization techniques that improves performance of object detection. Output is represented as a text file in CSV format that consists of frame numbers, positions and sizes of detected objects. Description of object classes is represented as a pretrained WaldBoost classifiers which contains LBP/LRD/LRD/CSLBP patterns.

Location

https://www.fit.vut.cz/research/product/584/

Projects

V3C - Visual Computing Competence Center, TAČR, Centra kompetence, TE01020415, 2012-2019, completed

Research groups

Computer Graphics Research Group (RG GRAPH)

Departments

Department of Computer Graphics and Multimedia (DCGM)