jsll / position-contrained-6dof-graspnet

This repository provides the code for training the position constrained generative grasp sampler from the paper Constrained Generative Sampling of 6-DoF Grasps
MIT License
12 stars 0 forks source link

Cannot begin training #4

Open shukla-yash opened 2 months ago

shukla-yash commented 2 months ago

Hi,

I followed your steps to set up the environment and download the dataset. I get this error when I try to start the training:

  File "position-contrained-6dof-graspnet/data/constrained_grasp_sampling_data.py", line 170, in load_from_pickle_file
    all_query_points_per_point_cloud = np.asarray(data["query_points/points_with_grasps_on_each_rendered_point_cloud"])
ValueError: setting an array element with a sequence. The requested array has an inhomogeneous shape after 1 dimensions. The detected shape was (100,) + inhomogeneous part.

data["query_points/points_with_grasps_on_each_rendered_point_cloud"] is a list and each element is of different size.

I tried passing it as a list but faced a problem later. Any suggestions will help! Thank you!