TEMOS: TExt to MOtionS
Generating diverse human motions from textual descriptions
Description
Official PyTorch implementation of the paper "TEMOS: Generating diverse human motions from textual descriptions".
Please visit our webpage for more details.
Bibtex
If you find this code useful in your research, please cite:
@article{petrovich22temos,
title = {{TEMOS}: Generating diverse human motions from textual descriptions},
author = {Petrovich, Mathis and Black, Michael J. and Varol, G{\"u}l},
journal = {arXiv},
month = {April},
year = {2022}
}
You can also put a star
Installation
👷
1. Create conda environment
conda create python=3.9 --name temos
conda activate temos
Install PyTorch 1.10 inside the conda environnement, and install the following packages:
pip install pytorch_lightning --upgrade
pip install torchmetrics==0.7
pip install hydra-core --upgrade
pip install hydra_colorlog --upgrade
pip install shortuuid
pip install tqdm
pip install pandas
pip install transformers
pip install psutil
pip install einops
The code was tested on Python 3.9.7 and PyTorch 1.10.0.
2. Download the datasets
KIT Motion-Language dataset
Be sure to read and follow their license agreements, and cite accordingly.
Use the code from Ghosh et al. or JL2P to download and prepare the kit dataset (extraction of xyz joints coodinates data from axis-angle Master Motor Map). Move or copy all the files which ends with "_meta.json", "_annotations.json" and "_fke.csv" inside the datasets/kit folder. "
AMASS dataset
WIP: instructions to be released soon
3. Download text model dependencies
Download distilbert from Hugging Face
cd deps/
git lfs install
git clone https://huggingface.co/distilbert-base-uncased
cd ..
4. SMPL body model
WIP: instructions to be released soon
5. (Optional) Donwload pre-trained models
WIP: instructions to be released soon
How to train TEMOS
🚀
The command to launch a training experiment is the folowing:
python train.py [OPTIONS]
The parsing is done by using the powerful Hydra library. You can override anything in the configuration by passing arguments like foo=value or foo.bar=value.
Experiment path
Each training will create a unique output directory (referred to as FOLDER below), where logs, configuations and checkpoints are stored.
By default it is defined as outputs/${data.dataname}/${experiment}/${run_id} with data.dataname the name of the dataset (see examples below), experiment=baseline and run_id a 8 unique random alpha-numeric identifier for the run (everything can be overridden if needed).
This folder is printed during logging, it should look like outputs/kit-mmm-xyz/baseline/3gn7h7v6/.
Some optional parameters
Datasets
data=kit-mmm-xyz: KIT-ML motions processed by the MMM framework (as in the original data) loaded as xyz joint coordinates (after axis-angle transformation → xyz) (by default)data=kit-amass-rot: KIT-ML motions loaded as SMPL rotations and translations, from AMASS (processed with MoSh++)data=kit-amass-xyz: KIT-ML motions loaded as xyz joint coordinates, from AMASS (processed with MoSh++) after passing through a SMPL layer and regressing the correct joints.
Training
trainer=gpu: training with CUDA, on an automatically selected GPU (default)trainer=cpu: training on the CPU
How to generate motions with TEMOS
Dataset splits
To get results comparable to previous work, we use the same splits as in Language2Pose and Ghosh et al.. To be explicit, and not rely on random seeds, you can find the list of id-files in datasets/kit-splits/ (train/val/test).
When sampling Ghosh et al.'s motions with their code, I noticed that their dataloader is missing some sequences (see the discussion here). In order to compare all the methods with the same test set, we use the 520 sequences produced by Ghosh et al. code for the test set (instead of the 587 sequences). This split is refered as gtest (for "Ghosh test"). It is used per default in the sampling/evaluation/rendering code. You can change this set by specifying split=SPLIT in each command line.
You can also find in datasets/kit-splits/, the split used for the human-study (human-study) and the split used for the visuals of the paper (visu).
Sampling/generating motions
The command line to sample one motion per sequence is the following:
python sample.py folder=FOLDER [OPTIONS]
This command will create the folder FOLDER/samples/SPLIT and save the motions in the npy format.
Some optional parameters
mean=false: Take the mean value for the latent vector, instead of sampling (default is false)number_of_samples=X: GenerateXmotions (by default it generates only one)fact=X: Multiplies sigma byXduring sampling (1.0 by default, diversity can be increased whenfact>1)
Model trained on SMPL rotations
If your model has been trained with data=kit-amass-rot, it produces SMPL rotations and translations. In this case, you can specify the type of data you want to save after passing through the SMPL layer.
jointstype=mmm: Generate xyz joints compatible with the MMM bodies (by default). This gives skeletons comparable todata=kit-mmm-xyz(needed for evaluation).jointstype=vertices: Generate human body meshes (needed for rendering).
Evaluating TEMOS (and prior works)
To evaluate TEMOS on the metrics defined in the paper, you must generate motions first (see above), and then run:
python evaluate.py folder=FOLDER [OPTIONS]
This will compute and store the metrics in the file FOLDER/samples/metrics_SPLIT in a yaml format.
Some optional parameters
Same parameters as in sample.py, it will choose the right directories for you. In the case of evaluating with number_of_samples>1, the script will compute two metrics metrics_gtest_multi_avg (the average of single metrics) and metrics_gtest_multi_best (chosing the best output for each motion). Please check the paper for more details.
Model trained on SMPL rotations
Currently, evaluation is only implemented on skeletons with MMM format. You must therefore use jointstype=mmm during sampling.
Evaluating prior works
WIP: the proper instructions and code will be available soon.
To give an overview:
- Generate motions with their code (it is still in the rifke feature space)
- Save them in xyz format (I "hack" their render script, to save them in xyz npy format instead of rendering)
- Load them into the evaluation code (instead of loading TEMOS motions).
Rendering motions
To get the visuals of the paper, I use Blender 2.93. The setup is not trivial (installation + running), I do my best to explain the process but don't hesitate to tell me if you have a problem.
Instalation
The goal is to be able to install blender so that it can be used with python scripts (so we can use ``import bpy''). There seem to be many different ways to do this, I will explain the one I use and understand (feel free to use other methods or suggest an easier way). The installation of Blender will be done as a standalone package. To use my scripts, we will run blender in the background, and the python executable in blender will run the script.
In any case, after the installation, please do step 5. to install the dependencies in the python environment.
- Please follow the instructions to install blender 2.93 on your operating system. Please install exactly this version.
- Locate the blender executable if it is not in your path. For the following commands, please replace
blenderwith the path to your executable (or create a symbolic link or use an alias).- On Linux, it could be in
/usr/bin/blender(already in your path). - On macOS, it could be in
/Applications/Blender.app/Contents/MacOS/Blender(not in your path)
- On Linux, it could be in
- Check that the correct version is installed:
blender --background --versionshould return "Blender 2.93.X".blender --background --python-expr "import sys; print('\nThe version of python is '+sys.version.split(' ')[0])"should return "3.9.X".
- Locate the python installation used by blender the following line. I will refer to this path as
path/to/blender/python.
blender --background --python-expr "import sys; import os; print('\nThe path to the installation of python of blender can be:'); print('\n'.join(['- '+x for x in sys.path if 'python' in (file:=os.path.split(x)[-1]) and not file.endswith('.zip')]))"
- Install these packages in the python environnement of blender:
path/to/blender/python -m pip install --user numpy
path/to/blender/python -m pip install --user matplotlib
path/to/blender/python -m pip install --user hydra-core --upgrade
path/to/blender/python -m pip install --user hydra_colorlog --upgrade
path/to/blender/python -m pip install --user moviepy
path/to/blender/python -m pip install --user shortuuid
Launch a python script (with arguments) with blender
Now that blender is installed, if we want to run the script script.py with the blender API (the bpy module), we can use:
blender --background --python script.py
If you need to add additional arguments, this will probably fail (as blender will interpret the arguments). Please use the double dash -- to tell blender to ignore the rest of the command. I then only parse the last part of the command (check temos/launch/blender.py if you are interested).
Rendering one sample
To render only one motion, please use this command line:
blender --background --python render.py -- npy=PATH_TO_DATA.npy [OPTIONS]
Rendering all the data
Please use this command line to render all the data of a split (which have to be already generated with sample.py). I suggest to use split=visu to render only a small subset.
blender --background --python render.py -- folder=FOLDER [OPTIONS]
SMPL bodies
Don't forget to generate the data with the option jointstype=vertices before. The renderer will automatically detect whether the movement is a sequence of joints or meshes.
Some optional parameters
downsample=true: Render only 1 frame every 8 frames, to speed up rendering (by default)canonicalize=true: Make sure the first pose is oriented canonically (by translating and rotating the entire sequence) (by default)mode=XXX: Choose the rendering mode (default ismode=sequence)video: Render all the frames and generate a video (as in the supplementary video)sequence: Render a single frame, withnum=8bodies (sampled equally, as in the figures of the paper)frame: Render a single frame, at a specific point in time (exact_frame=0.5, generates the frame at about 50% of the video)
quality=false: Render to a higher resolution and denoise the output (default to false to speed up))
License
This code is distributed under an MIT LICENSE.
Note that our code depends on other libraries, including SMPL, SMPL-X, PyTorch3D, Hugging Face, Hydra, and uses datasets which each have their own respective licenses that must also be followed.
364 Dec 14, 2022
111 Dec 31, 2022
35 Dec 6, 2022
307 Jan 3, 2023
153 Dec 14, 2022
1k Dec 31, 2022
574 Jan 2, 2023
115 Dec 28, 2022
49 Dec 1, 2022
284 Dec 21, 2022
24 Dec 21, 2022
100 Dec 18, 2022
1.5k Dec 28, 2022
47 Nov 6, 2022
19 May 4, 2022
23 Nov 5, 2022
58 Dec 31, 2022