Mathis Petrovich

Hello @jinfagang, I just release the files needed to render the motions (as in the figures of the paper). You can now use render.py after using sample.py (see more explanation...

Hello @jinfagang, Pre-trained models and instructions are now available. Please check the README.md.

Hello @L190201301, I will put more information in the README, in the next few weeks. (I may also distributes their motions.) To tell you what I use for comparison with...

Hello, That's a very good question. Actually, I am not doing something ideal, we can discuss a bit if you think of something better. After generating motions (from any method),...

Hello @Lucky-Maximize, I just solved this "pickle problem" with this commit: https://github.com/Mathux/TEMOS/commit/3e0efdd8acc253e7144a819c98e755b9e5c4dc1e Actually, the current code is not designed for GPU-training and it is not working yet. When I will...

Hello, I have the exact same problem. I tried with a machine on Ubuntu 20.04.4 and with the following versions: - emacs 27 / node 17 - emacs 28 /...

Hello @mahdiabdollahpour, You don't necessarily need to create .pkl files. You need to create a python file in the [src/datasets/](https://github.com/Mathux/ACTOR/tree/master/src/datasets) folder that corresponds to your custom dataset. Then you take...

Hello, You just need to change the architecture name from transformer to gru of fc. So the modelname can be: - cvae_transformer_rc_rcxyz_kl - cvae_gru_rc_rcxyz_kl - cvae_fc_rc_rcxyz_kl

Hi @AkbarShah96, Disclaimer: if you are using in the wild custom prompts, it will probably fail to generate the desired motion. In fact, I have already created this script but...

As this part of the code is borrowed from language2pose repository, I duplicate the issue there: https://github.com/chahuja/language2pose/issues/22