We propose an online tracking algorithm that adaptively models target appearances based on an online gradient boosting decision tree. Our algorithm is particularly useful for non-rigid and/or articulated objects since it handles various deformations of the target effectively by integrating a classifier operating on individual patches and provides segmentation masks of the target as final results. The posterior of the target state is propagated over time by particle filtering, where the likelihood is computed based mainly on a patch-level confidence map associated with a latent target state corresponding to each sample. Once tracking is completed in each frame, our gradient boosting decision tree is updated to adapt new data in a recursive manner. For effective evaluation of segmentation-based tracking algorithms, we construct a new ground-truth that contains pixellevel annotation of segmentation mask. We evaluate the performance of our tracking algorithm based on the measures for segmentation masks, where our algorithm illustrates outstanding accuracy compared to the state-of-the art segmentation-based tracking methods.


Figure 1. Success and precision plots in the non-rigid object tracking dataset


  title={Tracking-by-Segmentation with Online Gradient Boosting Decision Tree},
  author={Son, Jeany and Jung, Ilchae and Park, Kayoung and Han, Bohyung},
  booktitle={The IEEE International Conference on Computer Vision (ICCV)},


(Codes are not available at the moment, and will be released soon.)

Datasets and results

  • Non-rigid dataset with segmentation annotations (ZIP, 99.5 MB)

  • Our results (ZIP, 960 KB)

  • Bounding box results of other trackers (ZIP, 176 KB)

  • Segmentation results of other trackers (ZIP, 17.4 MB)