This folder contains data for training and testing the SRF. Example data is included here for illustration only. The full dataset should be downloaded from the project webpage:
http://www.umiacs.umd.edu/research/POETICON/geometric_affordance/. Follow the instructions in the enclosed README.txt for details on the folder structure. 

1. "Affordance_Part_Data/" contains "test" and "train" subfolders that follow the structure of the full dataset.

2. "Affordance_Part_Data_clutter/" is a placeholder for saving the 'cluttered' subset of the dataset.

The remaining four folders are precomputed data and have a flat structure:
3. "curvature" contains precomputed curvature values.

4. "normals" contains precomputed normal values.

5. "depth_clean" contains smoothed and cleaned-up depth values.

6. "label_rank" contains ranked ground truth labels.

More information on the dataset is available on the project webpage as well.

Dataset is maintained by Austin Myers [amyers-at-umiacs.umd.edu]

7. "DeepGraspingData" is where the Cornell grasping dataset, available at: http://pr.cs.cornell.edu/grasping/rect_data/data.php should be extracted into. You only need the /rawData subfolder of the dataset. 
