Predator: A Better Video Tracking Algorithm than Microsoft Kinect-from Czech doctoral thesis

Source: http://news.csdn.net/a/20110415/295938.html

Boycott (haha): I just saw this video to demonstrate a new object tracking algorithm. It is part of Zdenek kalal's doctoral thesis. Zdenek kalal is a Czech student at the University of surari. He demonstrated his magical precise positioning system, which can track almost any object in the camera as long as you can see it and select it. It can do a lot of look. In this video, he demonstrated shooting his fingers through the camera and selecting his fingers as the target. The system then can precisely track the movements of his fingers. What's even more surprising is that this system can improve tracking algorithms by analyzing the motion of objects. You can teach it to track your fingers, faces, or cars on the highway in a short time. With this system, we can almost implement man-machine interfaces like "Minority Report. Like Microsoft's Xbox's Kinect, this is better.

Kalal has 12 videos to demonstrate what he can do with this algorithm. As long as you have a good camera, you can install the software on your computer, tablet, or mobile phone, it can precisely locate a point on your forehead, your fingertip, or your eyes. When you put the camera out of the door, It can automatically identify the person you know or warn you that it is a stranger. People do not have to use their hands to easily manipulate computers. This technology has broad application prospects.

Http://v.youku.com/v_show/id_xmju5nzmxmjy#html youku video

You can find the code for this program from the University of surari website. It is free of charge. Kalal was awarded the technology everywhere scholarship. Http://info.ee.surrey.ac.uk/Personal/Z.Kalal/tld.html

Wonderful video demonstration: Click

Product introduction key features

• ● Input: video stream from single Monocular Camera, bounding box defining the object
• ● Output: object location in the stream, Object Model
• ● Implementation: MATLAB + C, single thread, no GPU
• ● No offline training stage
• ● Real-time performance on qvga video stream
• ● Dependence on opencv Library (single function)
• ● Ported to Windows, Mac OS X and Linux
• ● Illumination Invariant Features

Free Version (free source code can be downloaded from GitHub)

TLD can be downloaded for testing in a chosen application. we provide aprecompiled demo (Windows) and a source code that is released under GPL version 3.0. in short, it means that any distributed project that has des or links any portion of TLD source code has to be released with the source code under the GPL version 3.0 License or later.

Commercial Version

A license has to be purchased for Using TLD in a specified cial project. the licencing is managed by the IP owner, the University of Surrey and the operation is subject to negotiation. please contact the University of Surrey for further information.

More information

• ● High-level description of TLD
• ● Components of TLD
• ● Learning component of TLD
• ● Application of TLD tracker to faces
• ● Detailed description is in the following papers: iccv '09 (W), cvpr '10, ICIP '10, ICPR '10
• ● Explain technical questions (e.g. Installation) are being discussed in the followingdiscussion group.

Mitbbs comments

Lalaphin (Orpheus): MIT has been studying this stuff. As I have already said, Microsoft's ultimate goal in Using Kinect is not the gaming industry, but the exploration of Windows applications in the future.

Kz80.

Zlike (Final Fantasy): I think this does not solve the most critical problem: what to track? One of the difficulties of Object Tracking/recognition is how to determine what changes have taken place (because all pixels have changed) in an image that appears out of thin air ), instead of tracking a selected small area. If he does not have that selection step, but can directly identify the moving "valid" object, then he can be comparable to Kinnect.