This is the official implementation of the paper Generating Emotive Gaits for Virtual Agents Using Affect-Based Autoregression. Please use the following citation if you find our work uesful:
@inproceedings{bhattacharya2020generating,
author = {Uttaran Bhattacharya and Nicholas Rewkowski and Pooja Guhan and Niall L. Williams and Trisha Mittal and Aniket Bera and Dinesh Manocha},
title = {Generating Emotive Gaits for Virtual Agents Using Affect-Based Autoregression},
booktitle = {2020 {IEEE} International Symposium on Mixed and Augmented Reality, {ISMAR} 2020, November 9-13, 2020},
publisher = {{IEEE}},
year = {2020}
}
Our scripts have been tested on Ubuntu 18.04 LTS with
- Python 3.7
- Cuda 10.2
- cudNN 7.6.5
- PyTorch 1.5
- Unzip this repository.
We use $BASE to refer to the base directory for this project (the directory containing main.py). Change present working directory to $BASE.
- [Optional but recommended] Create a conda envrionment for the project and activate it.
conda create gen-emo-gaits-env python=3.7
conda activate gen-emo-gaits-env
- Install the package requirements.
pip install -r requirements.txt
Note: You might need to manually uninstall and reinstall matplotlib and kiwisolver for them to work.
- Install PyTorch following the official instructions.
Note: You might need to manually uninstall and reinstall
numpyfortorchto work.
We downloaded the Edinburgh Locmotion MOCAP Database, and annotated the locomotion data with emotion labels. Our processed and annotated dataset is available for download here: https://drive.google.com/file/d/1-D_tT09a7CfO3JSSHWX7-Ao3SjCxTsD2/view?usp=sharing. Unzip the downloaded file to a directorty named "data", located at the same level $BASE (i.e., $BASE and the data are sibling directories).
Run the main.py file with the appropriate command line arguments.
python main.py <args list>
The full list of arguments is available inside main.py.
For any argument not specificed in the command line, the code uses the default value for that argument.
On running main.py, the code will train the network and generate sample gestures post-training.
We also provide a pretrained model inside the models directory, available for download here: https://drive.google.com/file/d/1u8Ed_iPXdc-8bVzyRqXOfQ5g2QogNOj1/view?usp=sharing. Save the models directory directory under $BASE.
Set the command-line argument --train to False to skip training and use this model directly for evaluation. The generated samples are stored in the automatically created render directory. We generate test samples by deafult and also store the corresponding ground truth samples for comparison. We have tested that the samples, stored in .bvh files, are compatible with blender.