Our ECCV2024 workshop SpatialAI opens a localization challenge. We provide three different scenarios, each consisting of a reference map and a query sequence from the 4Seasons dataset
| scenarios | reference map sequence | query sequence |
|---|---|---|
| neighborhood | recording_2020-10-07_14-53-52 |
recording_2021-05-10_18-26-26 |
| business_campus | recording_2020-10-08_09-30-57 |
recording_2021-01-07_13-03-56 |
| countryside | recording_2020-06-12_11-26-43 |
recording_2021-01-07_14-03-57 |
Each reference map contains highly accurate reference poses. All sequences from the 4Seasons dataset can be used as training data.
-
recording_2020-10-07_14-53-52
-
recording_2020-10-08_09-30-57
-
recording_2020-06-12_11-26-43
-
recording_2021-05-10_18-26-26
-
recording_2021-01-07_13-03-56
-
recording_2021-01-07_14-03-57
For details on the dataset structure, we refer to the following documentation.
In this repository, you will find a set of tools for working with the 4Seasons dataset.
Each relocalizationFile_recording_source_to_recording_target.txt contains a list of first the keyframe from the source (reference) sequence and the query keyframe.
The task is to provide the relative pose between the source and the query frame.
Note that the relative pose is from cam0 of the reference sequence to cam0 of the query sequence, respectively.
The 6DOF poses are specified as translation (t_x, t_y, t_z), and quaternion (q_x, q_y, q_z, q_w).
The ground truth is withheld for the test sequences. Please submit your results as a single zip file. The results for each re-localization file must be stored in a separate txt file in the archive's root folder. The file name must be exactly like the provided ones (case-sensitive).
The file format should be a text file containing one re-localization pair per line. Each line must contain 9 values:
source_kf_id target_kf_id t_x t_y t_z q_x q_y q_z q_w
To submit your results, you will have to send an email with a single zip file which contains the re-localization results for the 3 provided files. Send to: spatialai.eccv2024@gmail.com.
Challenge deadline: Extended to 24:00 (CEST) Nov. 30, 2024
For questions, please send an email to: spatialai.eccv2024@gmail.com.