This is the first DSPS student competition on the application of AI for pavement condition monitoring. The competition will follow a data-centric model instead of the traditional model-centric approaches. Top-down views of pavement image data containing 7 main distress types annotated with bounding boxes and polygons will be provided. Participants will systematically change/enhance datasets provided using various data cleaning, annotation, augmentation strategies to improve the accuracy of a predefined model architecture.
- Nov 29th, 2021: Team registration begins! Github page of the challenge opens.
- Dec 3rd, 2021: First batch of training Data is released.
- Dec 13th, 2021: Second batch of training Data is released. Use link below to download unannotated images: https://drive.google.com/file/d/1a6YFGXggJJPbnK5tT7OHgllF8kCeYIUf/view?usp=sharing
- Dec 17th, 2021: Test Data is released. https://drive.google.com/file/d/1TtDGMgTyrH6bclBf76ggZk28N9aalOzc/view?usp=sharing
- Dec 17th, 2021: Leaderboard website is open for results submission. https://dsps-1e998.web.app/
- Jan 6th, 2022: Tutorial on how to use benchmark model. View a recording of the tutorial here: https://umsystem.hosted.panopto.com/Panopto/Pages/Viewer.aspx?id=b11433db-ca42-44ab-af29-ae150160ccd8
- Jan 21st, 2022: The link for submitting the source code enabled.
- Jan 25th, 2022: Competition ends! Deadline for submitting the source code & the solutions.
- Jan 31st, 2022: 10 teams have been shortlisted for paper submission!
- March 22nd, 2022: Symposium begins!
We expect you to respect the spirit of the competition and do not cheat.
- There is no restriction on the size of the team.
- The use of external data is forbidden
- Pre-trained models other than the ones provided are not allowed in the competition
- Teams must only use the algorithm/model selected by the organizers.
- Please submit source code and trained model before the deadline. IPython Notebook is desirable for the source code submission. The organizers will verify the reproducibility of the algorithm before determining the final winner.
Prizes will be offered to the top 3 rankers after the successful presentation of their paper at the conference.
- [First prize winner]: Cash Prizes or Conference travel reimbursement
- [Second prize winner]: Cash Prizes or Conference travel reimbursement
- [Third prize winner]: Cash Prizes or Conference travel reimbursement
$ git clone -b main_dev https://github.com/UM-Titan/DSPS.git
$ cd DSPS
$ open and run dsps_main.ipnyb # google collaboratory is preffered