This is the website for the dataset introduced in the ICRA 2014 publication "A Large-Scale 3D Database of Object Instances." Specifically, for each of (currently) 125 objects, we provide:
- 600 12 megapixel images, sampling the viewing hemisphere
- 600 registered RGB-D point clouds from a Carmine 1.09 sensor
- Pose information for each of the above images and point clouds
- Segmentation masks for each of the above images (and segmented point clouds)
- Merged point clouds consisting of data from all 600 viewpoints
- Reconstructed meshes from the merged point clouds
Note that some objects, depending on their properties (e.g. transparency) may not have point clouds or meshes. We include them to enable others to develop methods to reconstruct point clouds and/or meshes.
We plan to continuously collect and upload objects and test scenes. As researchers use the data, we will list results and benchmarks here. If you have any results on the data that you would like listed here, please contact us (bigbird@lists.eecs.berkeley.edu).
