vsum_dsf

The implementation of the paper "Video Summarization using Deep Semantic Features" in ACCV'16 [arXiv]

How to set it up

git clone https://github.com/mayu-ot/vsum_dsf.git

Install dependency

You can install required python packages using conda:

conda env create -f vsum_dsf/environment.yml

Requirements:

• numpy=1.11
• scipy
• scikit learn
• chainer=2.0

Optional:

• scikit video ( for exporting video )

This code utilizes tools provided by M. Gygli et al. [1]. You can set it up by:

cd vsum_dsf
git clone https://github.com/gyglim/gm_submodular.git
cd gm_submodular
python setup.py install --user

[1] Gygli, Grabner & Van Gool. Video Summarization by Learning Submodular Mixtures of Objectives. CVPR 2015.

Download dataset and model parameters

To test the model in the paper, download a data.zip HERE and extract it in the folder vsum_dsf.

The demo performs video summarization on the SumMe dataset (project page).

You can download the dataset as:

cd data/summe
wget https://data.vision.ee.ethz.ch/cvl/SumMe/SumMe.zip
unzip SumMe.zip

Example

See the notebook or:

python script/summarize.py
python script/evaluate.py results/summe/smt_feat