Martin Sundermeyer

Do you run your training on a Nvidia 30XX GPU? In that case, you might try the environment posted in #19

Sorry, I don't have those anymore. Would just recommend you to retrain the model with the available data.

It's indeed missing and I will add it shortly. For a quick fix you can change this https://github.com/NVlabs/contact_graspnet/blob/f28013a1a9f593fafc7c6dc0ca022b63464800a0/contact_graspnet/data.py#L679 to ```python self._renderer.change_scene([cad_path], [cad_scale], [np.eye(4)]) ```

Hi @Ruangq , sorry, but this is not enough information. Users have successfully replicated results using this repository. If your GPU is from the 30XX series, you might need to...

Hi @jucamohedano , sorry for the late answer. I used a rather simple but efficient approach by loading all contact points to GPU memory as a `tf.constant`. However, the size...

Hi @ColaDrinkerZ, do you have a 30xx GPU? It is probably related to this issue #9 where you need to install another TF/CUDA version. Best, Martin

yes you need to follow all steps described in the Readme :-) https://github.com/NVlabs/contact_graspnet#download-data

This seems to be an issue with your environment. Try to import tensorflow and check the version. TF 2.2. certainly has the compat mode: https://www.tensorflow.org/versions/r2.2/api_docs/python/tf/compat/v1

First, you definitely need the meshes for training, please follow the instructions in the README. Concerning your error, how much RAM do you have on your machine? Could you monitor...

afaik, there is no plan to provide an official pytorch version