DTU Deep Learning Course Project :
Affordable Text to Video Diffused Generation: Empirical Study
(Hugo Herrero, Xiaofu Chen, Yueting Guan,Yaxin Luo)
Technical University of Denmark
Installation Instructions on setting up the project environment:
1.Please follow the guidance in LAMP.ipynb for enviroment setup, training and inference.
2.(Warning for Training) At least 9 hours for a single A100 GPU training one motion pattern.
Evaluation Guidance on evaluating the project's performance:
1.Navigate to the Evaluation directory.
2.Follow instructions in Assesment.ipynb and other relevant scripts for assessment and analysis.
Dataset
-
The Training Videos are included in training_videos folder
-
images in benchmark folder could be used for inference as the first frame image with prompt.