Discover and Learn New Objects from Documentaries

Kai Chen   Hang Song   Chen Change Loy   Dahua Lin
Computer Vision and Pattern Recognition (CVPR) 2017, Honolulu, Hawaii

Abstract

Despite the remarkable progress in recent years, detecting objects in a new context remains a challenging task. Detectors learned from a public dataset can only work with a fixed list of categories, while training from scratch usually requires a large amount of training data with detailed annotations. This work aims to explore a novel approach -- learning object detectors from documentary films in a weakly supervised manner. This is inspired by the observation that documentaries often provide dedicated exposition of certain object categories, where visual presentations are aligned with subtitles. We believe that object detectors can be learned from such a rich source of information. Towards this goal, we develop a joint probabilistic framework, where individual pieces of information, including video frames and subtitles, are brought together via both visual and linguistic links. On top of this formulation, we further derive a weakly supervised learning algorithm, where object model learning and training set mining are unified in an optimization procedure. Experimental results on a real world dataset demonstrate that this is an effective approach to learning new object detectors.

Demo

Brief introduction

Demo

Dataset

The WLD Dataset is available at github.

Code

Code will be available at github.


Citation

@inproceedings{chen2017discover,
  author = {Kai Chen, Hang Song, Chen Change Loy, Dahua Lin},
  title = {Discover and Learn New Objects from Documentaries},
  booktitle = {Proceedings of IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  month = {July},
  year = {2017}
}

Contact

Kai Chen
ck015 [at] ie.cuhk.edu.hk