Home

Awesome

Dataset for End-to-end Affordance Learning

The dataset used in End-to-end Affordance Learning is based on Partnet Mobility datset.

We selected cabinet, chair and drawers from the original dataset for different tasks in our experiment.

We processed each object in the following manner:

Processed Dataset

You can download the processed dataset from Google Drive or manually build the dataset from original sapien Partnet Mobility dataset.

Manually Process Dataset

Since our dataset was based on Sapien Partnet Mobility, you need to first download the original dataset from Sapien, and put the dataset in 'asset' directory inside the root of your clone of this repo.

To reproduce the data preparation process, run create_xxx.py.

You may need to modify the 'root' variable to let the program access the downloaded dataset.