Objectron Versions Save

Objectron is a dataset of short, object-centric video clips. In addition, the videos also contain AR session metadata including camera poses, sparse point-clouds and planes. In each video, the camera moves around and above the object and captures it from different views. Each object is annotated with a 3D bounding box. The 3D bounding box describes the object’s position, orientation, and dimensions. The dataset contains about 15K annotated video clips and 4M annotated images in the following categories: bikes, books, bottles, cameras, cereal boxes, chairs, cups, laptops, and shoes

v0.1.1

2 years ago

The full set of models (for EfficientNet and MobilePose) are available for download at Objectron bucket. These models can be used to predict the 3D object poses from RGB images. Example usage, including Mobile, Python, and Web API are available via Mediapipe. To list/download the models, use gsutil ls gs://objectron/models

v0.1.0

3 years ago

Major Updates

Check our latest newsletter for more details.

  • Objectron Paper to appear in CVPR-2021.
  • Python and Web API for Objectron models (available via Mediapipe, Python colab).
    • Objectron models now offer a ready-to-use yet customizable Python solution as part of the prebuilt Mediapipe Python package.
    • The Mediapipe Python package is available on PyPI for Linux, macOS and Windows.
  • Camera pose refinement via offline bundle adjustment.
    • We updated the camera pose with more accurate poses obtained from running offline global bundle adjustment on the video.
    • The updated pose is very useful for applications that require more accurate and consistent camera poses, such as NeRF.
  • Community spotlight: Objectron is available via ActiveLoop hub.