YCB Object and Model Set: 3D Meshes and Images for Robotics Benchmarking

Over 80 physical objects are represented by mesh models, RGB, RGB-D, and point cloud images. The dataset was created by Yale University, Carnegie Mellon University, and UC Berkeley using two scanning systems: UC Berkeley's rig and the Google scanner. It includes high-resolution images and meshes at multiple polygon resolutions for each object.

Use Cases

• Benchmarking robotic grasping and manipulation algorithms based on provided 3D mesh models and Kinbody files.
• Training computer vision models for object recognition based on the 600 high-resolution RGB images per object.
• Developing 3D scene reconstruction methods based on the provided RGB-D images and point clouds.
• Simulating robotic interactions in environments like OpenRAVE using the provided Kinbody files.

Strengths

• Includes data for over 80 distinct physical objects.
• Provides 600 high-resolution RGB images, 600 RGB-D images, and 600 point cloud images per object from one scanning system.
• Offers 3D meshes at three different resolutions (16k, 64k, and 512k polygons) from a second scanning system.
• Data is collected by two state-of-the-art systems, suggesting high-quality capture.

Limitations

• Row count and total dataset size are unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.

Provenance

Source

Yale University, Carnegie Mellon University, and UC Berkeley.

Collection Method

Data collected by UC Berkeley's scanning rig and the Google scanner.

Time Range

null

Freshness

Last updated: unknown

Geography

null