Falling Things Dataset is a synthetic dataset presented by NVIDIA.
The dataset contains 60k annotated photos of 21 household objects taken from the YCB dataset.For each image, it contains the 3D poses, per-pixel class segmentation, and 2D/3D bounding box coordinates for all objects.The dataset provide mono and stereo RGB images, along with registered dense depth images.
Dataset can be downloaded at the following website: https://research.nvidia.com/publication/2018-06_falling-things-synthetic-dataset-3d-object-detection-and-pose-estimation
The directory structure should be:
FallingThings
├── fat
| ├── mixed
| | ├── kitchen_0
| | ├── kitchen_1
| | ├── kitchen_2
...
| ├── single
| | ├── 002_master_chef_can_16k
| | ├── 003_cracker_box_16k
...
Optionally you can write your own txt file and use all the parts of the dataset.
@inproceedings{tremblay2018falling,
title={Falling things: A synthetic dataset for 3d object detection and pose estimation},
author={Tremblay, Jonathan and To, Thang and Birchfield, Stan},
booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops},
pages={2038--2041},
year={2018}
}