Predator Object Tracking

Figuring out how to track objects in a video stream is a solution for many problems. There are a variety of applications waiting for more sophisticated software in this area. Most objects change in many ways as they move through the timespace of frames in video. They can move, they can rotate and the qualities of light reflecting from them changes. Many factors can contribute to high levels of complexity in attacking this problem.

Predator: A Smart Camera that Learns – [youtube.com]

Zdenek Kalal – [ee.surrey.ac.uk]

I am a researcher in computer vision, aiming to enable machines to “see”. Recently, I have defended my PhD at the University of Surrey, UK. My PhD advisors were Krystian Mikolajczyk and Jiri Matas. My external PhD examiner was Andrew Fitzgibbon from Microsoft Research Cambridge (Principal Researcher who worked on Kinect). As of January 2011, I am working at EPFL together with Vincent Lepetit and Pascal Fua on an exciting project for novel augmented reality applications.

In short, I am interested in self-improving real-time vision algorithms that are just on the boundary between blue sky research and industrial applications.

Tracking-Learning-Detection (TLD) – [ee.surrey.ac.uk]

TLD (aka Predator) is an algorithm for tracking of unknown objects in unconstrained video streams. The object of interest is defined by a bounding box in a single frame. TLD simultaneously tracks the object, learns its appearance and detects it whenever it appears in the video. The result is a real-time tracking that typically improves over time.

Tracking Videos – [youtube.com]
ekalic2’s Channel

Making Life Easier with Augmented Reality – [youtube.com]

Although the correct description “mobile markerless tracking algorithm optimized for dual-core chips” sounds very complex, the demo shows how easy life can be with AR. For example by getting virtual service instructions of a printer embedded into the live video.

Real-Time Markerless 3D Tracking – [youtube.com]

Edge-Based Markerless 3D Tracking of Rigid Objects using multiple hypotheses.

SEE ALSO:
…other articles on Augmented Reality

Comments are closed.