We provide a reimplementation of our CVPR'15 paper In Defense of Color-based Model-free Tracking.
In this work, we address color-based online object tracking where neither class-specific prior knowledge nor pre-learned object models are available. Combining a discriminative object-vs-background model with an additional distractor-aware term, we show that tracker based on standard color representations (such as histograms) can achieve state-of-the-art performance on a variety of test sequences.
Code & demo data (~5 MB, MATLAB)
If you use this tracker, please cite our paper:
BibTeX reference for convenience: