Tracking results of our adaptively boosted, gradient based tracker. It is not a specialized face tracker, but the appearance of the face is learned from first frame only.
This sequence especially demonstrates the adaption prevention in case of occlusions.
More information about the approach within the paper 'Boosting Scalable Gradient Features for Adaptive Real-Time Tracking', Klein and Cremers, ICRA 2011
Link to this comment:
All Comments (0)