`handbag_data/github_datasets/amazon-berkeley-objects/README.md`

13 lines
# Amazon Berkeley Objects (ABO)

Repository for additional ABO data and annotations used in:

<b>[ABO: Dataset and Benchmarks for Real-World 3D Object Understanding](https://arxiv.org/pdf/2110.06199.pdf)</b> <br>
Jasmine Collins, Shubham Goel, Kenan Deng, Achleshwar Luthra, Leon Xu, Erhan Gundogdu, Xi Zhang, Tomas F Yago Vicente, Thomas Dideriksen, Himanshu Arora, Matthieu Guillaumin, Jitendra Malik <br>
Conference on Computer Vision and Pattern Recognition (CVPR), 2022

Data and metadata used for <b>material prediction</b> and <b>multi-view retrieval</b> experiments can be found on the primary dataset [website](https://amazon-berkeley-objects.s3.amazonaws.com/index.html). Data, metadata and code used for <b>single-view 3D reconstruction</b> experiments can be found here.

## Download 3D renders dataset
The rendered ABO images + camera poses for 3D reconstruction benchmarking can be downloaded [here](https://amazon-berkeley-objects.s3.us-east-1.amazonaws.com/archives/abo-release-renders.zip) (223 GB). The dataset contains Blender renderings of 30 viewpoints for each of the 7,953 3D objects in ABO, as well as camera intrinsics and extrinsics for each rendering.