search

nutonomy / nuscenes-devkit

The devkit of the nuScenes dataset.
https://www.nuScenes.org
Other
2.3k stars 630 forks source link

Problem with NuScenes format to KITTI format #1071

Closed FrancoArtale closed 7 months ago

FrancoArtale commented 7 months ago

Hi!!

I'm using the export_kitti.py file, and when i used it with the v1.0-mini dataset of nuscenes i have a problem. In KITTI format we have the next:

<object_type, truncation, occlusion, alpha, left, top, right, bottom, height, width, length, x, y, z, rotation_y> --> 15 elements

and examples of the output of the export_kitti.py are: ['barrier', '0.00', '4', '-10.00', '0.00', '519.72', '58.90', '575.55', '1.1', '1.64', '0.36', '-17.09', '2.14', '26.36', '0.45', '0.0000'], ['traffic_cone', '0.00', '4', '-10.00', '49.28', '521.53', '65.74', '566.71', '0.96', '0.20', '0.21', '-16.70', '2.09', '27.20', '2.07','0.0000'], ['car', '0.00', '4', '-10.00', '889.41', '439.58', '944.63', '493.10', '3.1', '3.24', '7.10', '5.73','1.32', '79.10','-1.49','0.0000'],

16 elements in each. 4f545737bf3347fbbc9af60b0be9a963 4f545737bf3347fbbc9af60b0be9a963.txt

What's the problem here? What does not correspond?

export_kitti.py is in \nuscenes-devkit\python-sdk\nuscenes\scripts\export_kitti.py.

Greetings, FA.

whyekit-motional commented 7 months ago

@FrancoArtale the last element is the classification score for a given box: https://github.com/nutonomy/nuscenes-devkit/blob/4df2701feb3436ae49edaf70128488865a3f6ff9/python-sdk/nuscenes/utils/kitti.py#L410