contact_graspnet
contact_graspnet copied to clipboard
I have a question?
When I finish training, I use the generated weights for testing. After run 'python contact_graspnet/inference.py' , there is no grabbing gesture. I use cuda11.1 and tensorflow2.4.I want to know what causes this and how to solve this problem.
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 update the TF version. Have a look at the other issues.
I found a problem. In splits ,the content suffix is '.json' ,But the data type in the dataset is '.h5' .I wonder if this will affect the result.
But I found that this file was not used during the training.