uenian33 / semantic-cosegmentation Goto Github PK

• This repository includes the codes for our weakly-supervised video co-segmention. Please cite our paper if you use our code and model for your research.

• This code has been tested on Ubuntu 14.04 and MATLAB 2015a.

• Contact: Guangyu Zhong (guangyuzhonghikari at gmail dot com)

Semantic Co-segmentation in Videos. Yi-Hsuan Tsai*, Guangyu Zhong* and Ming-Hsuan Yang. European Conference on Computer Vision (ECCV), 2016. (* indicates equal contribution)

• Download and unzip the code.

• Install the attached caffe branch, as instructed at http://caffe.berkeleyvision.org/installation.html.

• Download the CNN model for feature extraction at http://vllab.ucmerced.edu/ytsai/CVPR16/pascal_segmentation.zip, then unzip the model folder under the caffe-cedn-dev/examples folder.

• Put your own videos in "Youtube_input" or another folder (you may need to change the corresponding paths).

• run demo_semantic_cosegment.m to generate tracklets.

• run demo_tracklets_feature.m to extract features for each tracklet.

• run demo_tracklets_submodular.m to select and merge tracklets.

• Currently this package only contains the implementation of our weakly-supervised video co-segmentation part.

• Please cite our paper if you find this work is useful.

@inproceedings{tsai2016semantic,
  title={Semantic Co-segmentation in Videos},
  author={Tsai, Yi-Hsuan and Zhong, Guangyu and Yang, Ming-Hsuan},
  booktitle={European Conference on Computer Vision},
  year={2016},
}

• Step 0: generate tracklets.
• Step 1 & 2: feature extraction and co-selection

• Check more results in our supplementary video.