Visualization

Rendering with Tier4

If you want to visualize annotation results, Tier4 supports some rendering methods as below.

Rendering Scene

>>> scene_token = t4.scene[0].token
>>> t4.render_scene(scene_token)

Rendering Instance(s)

>>> instance_token = t4.instance[0].token
>>> t4.render_instance(instance_token)

You can also render multiple instances at once:

>>> instance_tokens = [inst.token for inst in t4.instance[:3]]
>>> t4.render_instance(instance_tokens)

Rendering PointCloud

>>> scene_token = t4.scene[0].token
>>> t4.render_pointcloud(scene_token)

Note

In case of you want to ignore camera distortion, please specify ignore_distortion=True.

>>> t4.render_pointcloud(scene_token, ignore_distortion=True)

Save Recording

You can save the rendering result as follows:

>>> t4.render_scene(scene_token, save_dir=<DIR_TO_SAVE>)

If you don't want to spawn the viewer, please specify show=False as below:

>>> t4.render_scene(scene_token, save_dir=<DIR_TO_SAVE>, show=False)

Rendering with RerunViewer

If you want to visualize your components, such as boxes that your ML-model estimated, RerunViewer allows you to visualize these components.
For details, please refer to the API references.

>>> from t4_devkit.viewer import RerunViewer
# You need to specify `cameras` if you want to 2D spaces
>>> viewer = RerunViewer(app_id, cameras=<CAMERA_NAMES:[str;N]>)
# Rendering 3D boxes
>>> viewer.render_box3ds(seconds, box3ds)
# Rendering 2D boxes
>>> viewer.render_box2ds(seconds, box2ds)

It allows us to render boxes by specifying elements of boxes directly.

# Rendering 3D boxes
>>> viewer.render_box3ds(seconds, centers, rotations, sizes, class_ids)
# Rendering 2D boxes
>>> viewer.render_box2ds(seconds, rois, class_ids)