simple track data format

1. Detection format

[bboxes]

[types]

[velos]

 

bboxes[scene][frame] : scene의 모든 frame에서 detected된 객체의 bbox 정보. 하나의 bbox는 다음과 같은 정보로 되어있다.

ex) [238.55057929441963, 913.5869261780623, 0.8555841283783657, 1.8513201475143433, 4.525969505310059, 1.6440339088439941, -0.04874833532498536, -0.014557442911135306, -0.004948530062713921, -0.9986927418926963, 0.882607102394104]

순서대로 [translation(3), size(3), rotation(4), score(1)] 

 

types[scene][frame] : 객체의 class 정보.

ex) ['car']

 

velos[scene][frame] : scene의 모든 frame에서 detected된 객체의 velocity 정보.

ex) [-0.018035163166663593, 0.0028812381995912263]

순서대로 vx, vy

 

2. Tracking format

submission {
    "meta": {
        "use_camera":   <bool>  -- Whether this submission uses camera data as an input.
        "use_lidar":    <bool>  -- Whether this submission uses lidar data as an input.
        "use_radar":    <bool>  -- Whether this submission uses radar data as an input.
        "use_map":      <bool>  -- Whether this submission uses map data as an input.
        "use_external": <bool>  -- Whether this submission uses external data as an input.
    },
    "results": {
        sample_token <str>: List[sample_result] -- Maps each sample_token to a list of sample_results.
    }
}

 

simple track results

"meta" : {'use_camera': False, 'use_lidar': True, 'use_radar': False, 'use_map': False, 'use_external': False}

"results" : [sample token] {'translation': [238.55057929441963, 913.5869261780623, 0.8555841283783657], 'size': [1.8513201475143433, 4.525969505310059, 1.6440339088439941], 'velocity': [0.0, 0.0], 'rotation': [0.04869157725400045, 0.0, 0.0, 0.9988138616902138], 'tracking_score': 0.882607102394104, 'sample_token': 'fd8420396768425eabec9bdddf7e64b6', 'tracking_id': 'car_0_0', 'tracking_name': 'car'}

 

sample token : identifies the sample/keyframe for which object are detected.

translation : Estimated bounding box location in meters in the global frame: center_x, center_y, center_z

size : Estimated bounding box size in meters: width, length, height.

rotation : Estimated bounding box orientation as quaternion in the global frame: w, x, y, z.

velocity : Estimated bounding box velocity in m/s in the global frame: vx, vy.

tracking_id : object id.

tracking_name : predicted class.

tracking_score : object prediction score for the class identified by tracking_name.