Skip to content

submission.md

Latest commit

 

History

History
118 lines (102 loc) · 5.65 KB

submission.md

File metadata and controls

118 lines (102 loc) · 5.65 KB

Submission

Guide

Guide for CVPR 2024 AGC

🔥 CVPR 2024 AGC Track Mapless Driving

Driving Scene Topology

The submitted results of OpenLane-V2 subset-A test are required to be stored in a binary pickle file, which is a dict of identifier and formatted Map Element Bucket predictions a frame.

{
    'method':                               <str> -- name of the method
    'team':                                 <str> -- name of the team, identical to the Google Form
    'authors':                              <list> -- list of str, authors
    'e-mail':                               <str> -- e-mail address
    'institution / company':                <str> -- institution or company
    'country / region':                     <str> -- country or region, checked by iso3166*
    'results': {
        [identifier]: {                     <tuple> -- identifier of the frame, (split, segment_id, timestamp)
            'predictions': {
                'lane_segment':[            (i lane segments in the current frame)
                    {   
                        'id':               <int> -- unique ID in the current frame
                        'centerline':       <float> [n, 3] -- 3D coordiate
                        'left_laneline':    <float> [n, 3] -- 3D coordiate
                        'right_laneline':   <float> [n, 3] -- 3D coordiate
                        'confidence':       <float> -- confidence
                    },
                    ...
                ],
                'traffic_element':[         (j traffic elements in the current frame)
                    {
                        'id':               <int> -- unique ID in the current frame
                        'attribute':        <int> -- attribute of traffic element
                        'points':           <float> [2, 2] -- top-left and bottom-right corners of the 2D bounding box
                        'confidence':       <float> -- confidence, only for prediction
                    },
                    ...
                ],
                'area':[                    (k areas in the current frame)
                    {   
                        'id':               <int> -- unique ID in the current frame
                        'category':         <int> -- area category
                        'points':           <float> [n, 3] -- 3D coordiate
                        'confidence':       <float> -- confidence, only for prediction
                    },
                    ...
                ],
                'topology_lsls':            <float> [n, n] -- adjacent matrix among lane segments
                'topology_lste':            <float> [n, k] -- adjacent matrix between lane segments and traffic elements
            }
        },
        ...
    }
}
  • The #points of lane segment and area could be various. But we recommend 10 for each line in lane segment and 20 for area to keep align with ground truth.
  • We recommend astype all float to np.float16 to reduce the submission file size.

Test before submission

Hugging Face server will not return any detailed error if submission failed. Please test the submission file of val set before submit.

from openlanev2.lanesegment.evaluation.evaluate import evaluate

metrics = evaluate(
    ground_truth='data/OpenLane-V2/data_dict_subset_A_val_ls.pkl', 
    predictions='/path/to/submission.pkl'
)
print(metrics)

Steps

  1. Register for your team by filling in this Google Form.
  2. Prepare your results formatted as mentioned above.
  3. Follow the steps in the Submission Information tab of the competition space.

Submission Guide for CVPR 2023 Challenge

🔥 CVPR 2023 OpenLane Topology Challenge

OpenLane Topology

The submitted results are required to be stored in a pickle file, which is a dict of identifier and formatted predictions of a frame:

{
    'method':                               <str> -- name of the method
    'authors':                              <list> -- list of str, authors
    'e-mail':                               <str> -- e-mail address
    'institution / company':                <str> -- institution or company
    'country / region':                     <str> -- country or region, checked by iso3166*
    'results': {
        [identifier]: {                     <tuple> -- identifier of the frame, (split, segment_id, timestamp)
            'predictions': {
                'lane_centerline':              ...
                'traffic_element':              ...
                'topology_lclc':                ...
                'topology_lcte':                ...
            }
        },
        ...
    }
}

Steps

  1. Create a team on EvalAI.
  2. Click the 'Participate' tag, then choose a team for participation.
  3. Choose the phase 'Test Phase' and upload the file formatted as mentioned above.
  4. Check if the submitted file is valid, which is indicated by the 'Status' under the tag of 'My Submissions'. A valid submission would provide performance scores.