CIRCLE: Capture In Rich Contextual Environments (CVPR 2023)

This repo hosts the website for CIRCLE dataset.

Download

The CIRCLE data are hosted in an AWS S3 bucket. We provide the motion data separate from the first person videos. Please use the following links to download:

• Motion (SMPL-X and BVH) and headset trajectories
• Habitat first person videos
• Blender first person videos
• Scene and subject URDF files
  • Scene without doors

If you find any issues with the dataset, please let us know.

CIRCLE data are licensed under a CC BY-NC 4.0 license. For the license of the code in this repository, see the LICENSE file.

Visualization

The recommended way to visualize CIRCLE data is to use Blender. All sequences are stored assuming a Y up, -Z forward frame of reference. Below we describe the steps required to make sure the sequences are loaded in the correct orientation.

Visualizing using Blender.

Use the Blender import menu to import the scene .glb file. To visualize the motion, either use File -> Import -> Motion capture (.bvh) (make sure to enable Scale FPS and Update Scene Duration), or download and install the SMPL-X Blender add-on, and follow the instructions below:

1. On the 3D viewport, press N to show the sidebar.
2. Click on the SMPL-X separator.
3. Click Add Animation. Set the format to SMPL-X and load the .npz file of the sequence you would like to inspect.

Visualizing using Habitat.

We also provide a custom habitat-sim viewer, which exemplifies how to load CIRCLE sequences into Habitat (check the __init__ method of circle_viewer.py and the next_pose method in mocap_interface.py).

Rendering

To reproduce the first person videos distributed with CIRCLE, check the render folder.