how to find the corresponding camera type only through "sample_annotation"

[yushengjiexy]

I got the sample_annotation "1408af8e897d40fbbd78c33b59f8821f", then i render the annotation.


You can see the picture captured by "CAM_FRONT_LEFT".
I want to ask you how to get the camera type by code.

[yushengjiexy]

Because i want to convert "obj_pose_in_world" to "obj_pose_in_cam".
Thank you!

[whyekit-motional]

@yushengjiexy you could try something like this:

from typing import Optional
from nuscenes.nuscenes import NuScenes


def get_cam_of_sample_annotation_token(nusc: NuScenes, sample_ann_token: str) -> Optional[str]:
    """
    Figure out which camera the object is fully visible in.
    """
    sample_ann = nusc.get('sample_annotation', sample_ann_token)
    sample = nusc.get('sample', sample_ann['sample_token'])

    # Figure out which camera the object is fully visible in (this may return nothing).
    cams = [key for key in sample['data'].keys() if 'CAM' in key]
    for cam in cams:
        _, boxes, _ = nusc.get_sample_data(sample['data'][cam], selected_anntokens=[sample_ann_token])
        if len(boxes) > 0:
            return cam  # We found an image that matches. Let's abort.
        

nusc_ = NuScenes(version='v1.0-trainval', dataroot='/data/sets/nuscenes', verbose=False)

print(get_cam_of_sample_annotation_token(nusc_, '1408af8e897d40fbbd78c33b59f8821f'))

Note that this was modified from render_annotation