https://github.com/j-min/Adversarial_Video_Summary

Unofficial PyTorch Implementation of SUM-GAN from "Unsupervised Video Summarization with Adversarial LSTM Networks" (CVPR 2017)
https://github.com/j-min/Adversarial_Video_Summary

gan pytorch summarization vae video

Last synced: 7 days ago
JSON representation

Unofficial PyTorch Implementation of SUM-GAN from "Unsupervised Video Summarization with Adversarial LSTM Networks" (CVPR 2017)

• Host: GitHub
• URL: https://github.com/j-min/Adversarial_Video_Summary
• Owner: j-min
• Created: 2017-11-13T19:06:18.000Z (about 7 years ago)
• Default Branch: master
• Last Pushed: 2023-03-28T18:20:30.000Z (over 1 year ago)
• Last Synced: 2024-08-02T20:43:42.277Z (3 months ago)
• Topics: gan, pytorch, summarization, vae, video
• Language: Python
• Homepage:
• Size: 217 KB
• Stars: 239
• Watchers: 11
• Forks: 62
• Open Issues: 19
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Adversarial_Video_Summary

## PyTorch Implementation of SUM-GAN

- from ["Unsupervised Video Summarization with Adversarial LSTM Networks  (CVPR 2017)"](http://web.engr.oregonstate.edu/~sinisa/research/publications/cvpr17_summarization.pdf)

- by Behrooz Mahasseni, Michael Lam and Sinisa Todorovic

- Code Author: [Jaemin Cho](https://github.com/j-min)

- Used as baseline for unsupervised video summarization

## Changes from Original paper

- Video feature extractor

    - GoogleNet pool5 (1024) => ResNet-101 pool5 (2048)

    - Followed by linear projection to 500-dim

- Stable GAN Training

    - Discriminator's learning rate: 1e-5 (Others: 1e-4)

    - Fix Discriminators' parameters for first 15 steps at every epoch.

## Model figures

![Model figure 1](imgs/fig1.png)

![Model figure 2](imgs/fig2.png)

![Model figure 3](imgs/fig3.png)

## Algorithm

![Algorithm figure](imgs/fig4.png)