tlpss2025cc-by-4.0

GITW (Glasses-in-the-Wild), re-split for 2D-keypoints-benchmark

A repackaging of the Glasses-in-the-Wild dataset containing 1000 crowdsourced 256x256 images of transparent and partially filled drinking glasses with 5 keypoint annotations per image for keypoint detection tasks.

Downloads0
Episodes1000

Why This Matters for Physical AI

This dataset enables training of robust keypoint detection models on transparent objects, which is critical for robotic manipulation and vision systems that must handle reflective and non-opaque materials.

Technical Profile

Modalities
rgb
Environment
lab
Task Types
keypoint-detection
Episodes
1000
Annotation Types
keypointsbounding_boxes
License
cc-by-4.0
Part of the 2D-keypoints-benchmark family

Access

Need custom rgb data?

Claru builds purpose-built datasets for lab applications with dense human annotations and quality assurance.

Request a Sample Pack

Related Datasets