Communications About Ranking, Challenge Paper, and Workshop

Communications About Ranking, Challenge Paper, and Workshop  

  By: f.bolelli on Aug. 21, 2024, 12:02 p.m.

Dear participants,

It has been a pleasure seeing all of you compete in our challenge until the very last minute. Congratulations on having your algorithm submitted and evaluated for the final phase of ToothFairy2!

In the following, I will report important information about the next steps regarding the winner's announcement, the challenge resuming paper, and the call for papers for the ToothFairy workshop. Please read the post carefully until the end.

Let's Start with the Ranking

Although providing valuable insights, the final phase ranking available on grand-challenge is not definitive. Because of configuration constraints, the live leaderboard follows a simplified ranking schema, which averages over two ranks (mean Dice and HD95 on all classes). However, as specified in the challenge guidelines (structured challenge submission, ranking and prizes), the final ranking will be the average over 84 different rankings, one for each pair class-metric. If ties at the top of the leaderboard will remain, they will be broken considering algorithm execution time and memory requirements. Moreover, two docker containers failed the upload because of a grand-challenge connection issue on the evening of August 18th: they were shared with us on time and will be included and evaluated sooner using grand-challenge to ensure the fairest comparison. The winners will be officially communicated as soon as we have evaluated the top-ranked submissions' code and descriptive document. In this regard, I ask all the participants at the top of the leaderboard to share their code and the descriptive document ASAP (by email will be fine). The official deadline is August 31st, but the sooner, the better. A draft of the paper will be acceptable as well; it can be later refined to be submitted to our ToothFairy workshop!

About the Challenge Paper

We will start the writing process in a few weeks with the hope of having a first draft ready in Morocco! 💪As you may know, all the participants who submitted a valid solution other than our example code are eligible to be authors of that paper. In order to collect the necessary information, we kindly ask you to fill in the Google form available at https://forms.gle/nq8AGgu14cCnpFuP6. The deadline is August 31st!

About the ToothFairy Workshop

As you probably know, the ToothFairy challenge post-proceeding will be handled through Springer this year. We invite all of you to submit a paper describing your challenge submission through OpenReview (https://openreview.net/group?id=MICCAI.org/2024/Workshop/ToothFairy&referrer=%5BHomepage%5D(%2F)#tab-your-consoles). The deadline is October 1st. You can follow the Springer LNCS template for the writing process; workshop-specific guidelines will be released in the next few days through the grand-challenge portal. You can already find general guidelines at https://toothfairy2.grand-challenge.org/call-for-papers/.

Please note that all of the participants who aim to become authors of the joint paper and do not plan to submit their work to the ToothFairy workshop have to provide a document describing their submission!

Presenting your Work @ MICCAI

We would really like to have all of you presenting your work at the MICCAI on-site event (October 6th, Marrakech, Morocco), but the schedule is very tight. We will collect your availability in the following weeks and prioritize those placed higher in the ranking. Anyway, you are all more than welcome to attend the workshop and join our scientific discussion!

Cheers,
Federico Bolelli on behalf of the ToothFairy2 organizers.

 Last edited by: f.bolelli on Aug. 23, 2024, 8:13 a.m., edited 3 times in total.

Re: Communications About Ranking, Challenge Paper, and Workshop  

  By: f.bolelli on Aug. 23, 2024, 8:20 a.m.

Dear all,

In the above post (now updated), I failed to highlight that, because of configuration constraints, the live leaderboards follow a "simplified" ranking schema based on the average Dice and HD95 metrics on all classes. However, as detailed in the structured challenge submission (https://ditto.ing.unimore.it/static/toothfairy2/ToothFairy2_StructuredChallengeDescription.pdf) and reported in the grand-challenge page (https://toothfairy2.grand-challenge.org/ranking-prices/), the final ranking will be the average of 84 different rankings, one for each couple of class-metric, strictly following our original guidelines.

I apologize for any misunderstanding.

Best regards,
Federico Bolelli on behalf of ToothFairy2 challenge organizers.