Skip to content
Snippets Groups Projects
README.md 6.57 KiB
Newer Older
Hussain Kanafani's avatar
Hussain Kanafani committed
## Evaluating and Extending Unsupervised VideoSummarization Methods
Hussain Kanafani's avatar
Hussain Kanafani committed

Hussain Kanafani's avatar
Hussain Kanafani committed
1. Evaluation unsupervised methods with different metric under same configuration.
2. Investigating effect of extracted features in unsupervised methods and extend it to perform better than baseline work.
Hussain Kanafani's avatar
Hussain Kanafani committed
3. Identify gaps form last step and try to fill by porposed solution by extending or modifying an existing model.
Hussain Kanafani's avatar
Hussain Kanafani committed

Hussain Kanafani's avatar
Hussain Kanafani committed
**Abstract**

Hussain Kanafani's avatar
Hussain Kanafani committed
This study validates the recent work in unsupervised video summarization and extends the experiments with feature variation.
In  our  understanding,  deep learning based approaches select frame-candidates for a video summary.
These frames are drawn with a probability that can be used to calculate the scaled importance for comparison with the basic truth as an evaluation.
However, most works in the literature focus on network architecture and follow the same feature extraction technique using the resulting deep features from one pretrained model.
Hussain Kanafani's avatar
Hussain Kanafani committed
At this point a notable gap of the existing approaches is feature variation, which this study explores.  
Hussain Kanafani's avatar
Hussain Kanafani committed
Building on that, model variables like network architecture, optimizer, and activation functions can have an impact on the performance combined with different feature selection techniques that haven’t beenexplored yet.Further,  the Evaluation of existing models is conducted using one metric,which  may  be  not  representative  in  video  summarization  task,  since  it  could ignore key-frames in favor of other frames and still have a relatively high value. This work performs feature extraction using multiple pretrained neural models, and then measures the impact of them on current state-of-the-art works.Then,it evaluates the state-of-the-art works using different evaluation metrics.
Eventually, it aims to find an unsupervised video summarization method, in order to fill the gaps, and leverage the existing works.
Hussain Kanafani's avatar
Hussain Kanafani committed

Hussain Kanafani's avatar
Hussain Kanafani committed
**Important Wiki Pages:**

Hussain Kanafani's avatar
Hussain Kanafani committed
* [Notes on SUM-GAN-AAE (Apostolidis et al. 2020)](https://gitlab.uni-hannover.de/hussainkanafani/unsupervised-video-summarization/-/wikis/SUM-GAN-AAE-(Apostolidis-et-al.-2020))
Hussain Kanafani's avatar
Hussain Kanafani committed

* [References and important findings](https://gitlab.uni-hannover.de/hussainkanafani/unsupervised-video-summarization/-/wikis/Findings)

Hussain Kanafani's avatar
Hussain Kanafani committed
* [Reproduce (Zhou et al. 2018)](https://gitlab.uni-hannover.de/hussainkanafani/unsupervised-video-summarization/-/wikis/Reproduce-(Zhou-et-al.-2018)%E2%80%8B)
Hussain Kanafani's avatar
Hussain Kanafani committed

* [Conda cheat sheet](https://gitlab.uni-hannover.de/hussainkanafani/unsupervised-video-summarization/-/wikis/Conda-cheat-sheet)
Hussain Kanafani's avatar
Hussain Kanafani committed
# Datasets
Structured h5 files with the video features and annotations of the SumMe and TVSum datasets are available within the "data" folder. The GoogleNet features of the video frames were extracted by [Ke Zhang](https://github.com/kezhang-cs) and [Wei-Lun Chao] and the h5 files were obtained from [Kaiyang Zhou](https://github.com/KaiyangZhou/pytorch-vsumm-reinforce). 

These files have the following structure:
```
/key
    /features                 2D-array with shape (n_steps, feature-dimension)
    /gtscore                  1D-array with shape (n_steps), stores ground truth improtance score (used for training, e.g. regression loss)
    /user_summary             2D-array with shape (num_users, n_frames), each row is a binary vector (used for test)
    /change_points            2D-array with shape (num_segments, 2), each row stores indices of a segment
    /n_frame_per_seg          1D-array with shape (num_segments), indicates number of frames in each segment
    /n_frames                 number of frames in original video
    /picks                    positions of subsampled frames in original video
    /n_steps                  number of subsampled frames
    /gtsummary                1D-array with shape (n_steps), ground truth summary provided by user (used for training, e.g. maximum likelihood)
    /video_name (optional)    original video name, only available for SumMe dataset
```
Original videos and annotations for each dataset are also available in the authors' project webpages:

**TVSum dataset**: [https://github.com/yalesong/tvsum](https://github.com/yalesong/tvsum) 


**SumMe dataset**: [https://gyglim.github.io/me/vsum/index.html#benchmark](https://gyglim.github.io/me/vsum/index.html#benchmark)



### CSNet
We used the implementation of [SUM-GAN](https://github.com/j-min/Adversarial_Video_Summary) method as a starting point to implement CSNet.

#### How to train
The implementation of CSNet is located under the directory csnet. Run main.py file with the configurations specified in configs.py to train the model.



Hussain Kanafani's avatar
Hussain Kanafani committed
### SUM-Ind
Make splits
```bash
python create_split.py -d datasets/eccv16_dataset_summe_google_pool5.h5 --save-dir datasets --save-name summe_splits  --num-splits 5
```
As a result, the dataset is randomly split for 5 times, which are saved as json file.

Train and test codes are written in `main.py`. To see the detailed arguments, please do `python main.py -h`.

#### How to train
```bash
python main.py -d datasets/eccv16_dataset_summe_google_pool5.h5 -s datasets/summe_splits.json -m summe --gpu 0 --save-dir log/summe-split0 --split-id 0 --verbose
```

#### How to test
```bash
python main.py -d datasets/eccv16_dataset_summe_google_pool5.h5 -s datasets/summe_splits.json -m summe --gpu 0 --save-dir log/summe-split0 --split-id 0 --evaluate --resume path_to_your_model.pth.tar --verbose --save-results
```


Hussain Kanafani's avatar
Hussain Kanafani committed
### Citations
Hussain Kanafani's avatar
Hussain Kanafani committed
```
@article{zhou2017reinforcevsumm, 
   title={Deep Reinforcement Learning for Unsupervised Video Summarization with Diversity-Representativeness Reward},
   author={Zhou, Kaiyang and Qiao, Yu and Xiang, Tao}, 
   journal={arXiv:1801.00054}, 
   year={2017} 
}
Hussain Kanafani's avatar
Hussain Kanafani committed
```

```
@inproceedings{DBLP:conf/aaai/JungCKWK19,
  author    = {Yunjae Jung and
               Donghyeon Cho and
               Dahun Kim and
               Sanghyun Woo and
               In So Kweon},
  title     = {Discriminative Feature Learning for Unsupervised Video Summarization},
  booktitle = {The Thirty-Third {AAAI} Conference on Artificial Intelligence, {AAAI}
               2019, The Thirty-First Innovative Applications of Artificial Intelligence
               Conference, {IAAI} 2019, The Ninth {AAAI} Symposium on Educational
               Advances in Artificial Intelligence, {EAAI} 2019, Honolulu, Hawaii,
               USA, January 27 - February 1, 2019},
  pages     = {8537--8544},
  publisher = {{AAAI} Press},
  year      = {2019},
  url       = {https://doi.org/10.1609/aaai.v33i01.33018537},
  doi       = {10.1609/aaai.v33i01.33018537},
  timestamp = {Wed, 25 Sep 2019 11:05:09 +0200},
  biburl    = {https://dblp.org/rec/conf/aaai/JungCKWK19.bib},
  bibsource = {dblp computer science bibliography, https://dblp.org}
}
Hussain Kanafani's avatar
Hussain Kanafani committed
```