https://github.com/compvis/unsupervised-disentangling
https://github.com/compvis/unsupervised-disentangling
Last synced: about 1 year ago
JSON representation
- Host: GitHub
- URL: https://github.com/compvis/unsupervised-disentangling
- Owner: CompVis
- Created: 2019-03-15T12:22:03.000Z (about 7 years ago)
- Default Branch: master
- Last Pushed: 2020-10-17T15:35:33.000Z (over 5 years ago)
- Last Synced: 2025-03-21T12:07:28.897Z (about 1 year ago)
- Language: Python
- Size: 133 MB
- Stars: 55
- Watchers: 9
- Forks: 11
- Open Issues: 5
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
git README.md
Work in progress
Instructions:
Install tensorflow 1.14
Download
a) video dataset or,
b) image dataset
Change dataloader
a) use the structure of load_train_human3m in dataloading.py
b) use the structure of load_train_generic in dataloading.py
Start training (current hyperparameters work for roughly cropped human3m without background)
a) python main.py test_a --gpu 0 --dataset human3m --mode train --bn 16
b) python main.py test_b --gpu 0 --dataset generic --mode train --bn 16 --static
Eval (use same flags as in training except for mode)
a) python main.py test_a --gpu 0 --dataset human3m --mode predict --bn 16
b )python main.py test_b --gpu 0 --dataset generic --mode train --bn 16 --static
Todo:
-code documentation
-add preprocessing
-add hyperparameters for datasets
-add eval functions
[Project page with videos](https://compvis.github.io/unsupervised-disentangling/)