Object Recognition (Nike logo)
Posted on : 22-10-2009 | By : Alex | In : Demo, Demo video, Demo videos, OpenCV, YouTube
3
This is a demo video of the invariant orientation and scale fast object detection algorithm. The algorithm is a robust in cases when the object is deformed a little 
The algorithm is a cross-platform solution.
Performance:
- on ARM11 530MHz, the algorithm gives 1 fps for 640×480 frame;
- on Intel P4 3Hz, the algorithm gives 12 fps and more for 640×480 frame.
Quality: 86%.
what algorithm are you using to achieve the result? recognization Rotated NIKE logo is difficult. If not for rotation, haar + adaboost can achieve!
Hi YCat,
You are right, this is a difficult task. This is not haar + adaboost, we rejected Viola-Jones approach since it doesn’t work at all if an object is rotated. Nike-logo detection is original own algorithm. For nike-logo detection we used template matching, key-points matching and other and of course original own heuristics.
it is really a great job