projectaria/aria-digital-twin

Aria Digital Twin (ADT) Dataset

Figure 1: An overview of the ADT dataset

Dataset Summary

ADT provides raw and synthesized sensor data from Project Aria glasses, combined with groundtruth data generated using a motion capture system including depth images, device trajectories, object trajectories and bounding boxes, and human tracking. We also provide processed sensor data from our Machine Perception Services. Go to ADT Data Format to see a full list of the data we provide.

The ADT dataset contains 236 sequences recording single and dual-person activities. The data was recorded in two spaces: an apartment and a single room office. There are 74 single-instance dynamic objects shared between the two spaces.

Go to the projectaria.com/datasets/adt/ to learn more about the dataset, including instructions for downloading the data, installing our dataset specific tooling, and running example tutorials.

Dataset Contents

ADT contains 236 sequences, each sequence consists of a single Aria device recording (in VRS format) along with all ground truth data that accompanies this recording. When multiple devices were recording data concurrently, they are always time synchronized and all data collected in each scene is expresssed in the same World coordinate frame. Along with ground truth data, we also provide output from our Machine Perception Services (MPS) which you can learn more about here. All data included in each sequences is broken down in the table below:

Data type	What's included	Per sequence size	Total size for all sequences
main Aria	Aria raw data, 2D bounding box, 3D object poses and bounding box, skeleton data, eye gaze data	3 - 6 GB	~700 GB
segmentation	Instance segmentation data	2 - 4 GB	~750 GB
depth	Depth map data	4 - 8 GB	~1.5 TB
synthetic	Synthetic rendering data	2 - 4 GB	500 GB
MPS eyegaze	Eyegaze, summary file	< 1 MB	~31 MB
MPS SLAM points	Semi-dense points and observations	200 - 500 MB	~31 GB
MPS SLAM trajectories	Open and closed loop trajectories	100 - 200 MB	12 GB
MPS SLAM online calibration	Online calibrations	< 20 MB	1.2 GB

Citation Information

If using ADT, please cite our research paper which can be found here.

License

ADT license can be found here.

Contributors

@nickcharron