Dataset Card for Dropjects_Real

Dropjects_real is a stereo RGB-D object dataset captured at the Chair of Cyber-Physical Systems in Production Engineering at the Technical University of Munich. It contains pose, bounding box, and segmentation masks for a 9 objects organized in different scenes, clutter and lighting conditions.

Dataset Details

Subsets

You can choose any specific scene with the following pattern: "{class}-{scene}-{clutter}-{lighting}". You can also use the wildcard "{}", to choose all available options. Choose from the following:

CLASSES:

• battery_holder
• buckle_socket
• buckle_plug
• chew_toy_big
• cpsduck
• cpsglue_big
• group1 consists of cpsduck, battery_holder, chew_toy_big, cpsglue_big, stapler
• group2 consists of buckle_plug, buckle_socket, nema17_holder, nema17_holder2
• nema_holder consists of nema17_holder, nema17_holder2
• stapler

SCENES:

• box
• array only for group1 and group2, and only uncluttered

CLUTTER:

• cluttered
• uncluttered

LIGHTING:

• diffuseRight
• diffuseLeft
• spotRight
• spotLeft
• dark
• normal

Then you can load the dataset like this, for example all lighting conditions for the stapler in the box, with clutter

from datasets import load_dataset

ds = load_dataset("LukasDb/dropjects_real", "stapler-box-cluttered-{}", streaming=True, trust_remote_code=True, split='test')
for data in ds.with_format('tensorflow'):
    rgb = data['rgb'] # tf.uint8 Tensor, (h,w,3)

Dataset Sources

• Repository: TBA
• Paper: TBA

Dataset Structure

TBA

Citation

BibTeX: TBA

Dataset Card Authors and Contact

Lukas Dirnberger (lukas.dirnberger@tum.de)