Grasping Dataset

The data here is from following publication:

Learning Hand-Eye Coordination for Robotic Grasping with Deep Learning and Large-Scale Data Collection

Sergey Levine, Peter Pastor, Alex Krizhevsky, Deirdre Quillen

ISER, 2016. [blog post] [arXiv] [PDF]

These files are no longer available.

This data set contains roughly 650,000 examples of robot grasping attempts. The examples are grouped into batches that have identical feature keys. The set of features present in each batch is described in a CSV file.

• • Feature descriptions and counts -- This sheet describes the meanings of the features in the grasping dataset and shows which batches contain which features.

The Example protobufs are stored in Tensorflow RecordWriter format. The files are named:

grasping_dataset_<batch>.tfrecord-<k>-of-<n>

The number in the <batch> string is the number of features in each example. Batches that are larger than 500MB are split into parts.

The script download_listing.sh linked at the bottom of the page will use curl to download all (886GB) of the data files.

$ ./download_listing.sh grasp_listing.txt ./grasping_dataset

The file grasp_listing_hash.txt contains a sha256 hash value for each file. (Thank you to Andrew Hundt.)

License: This data is licensed by Google Inc. under a Creative Commons Attribution 4.0 International License.