EndoCV2022 Challenge Rounds:¶

EndoCV2022 is aimed at promoting "novel DL method development in endoscopy" and reduce the attempts to participate to win concept with repetitively same method at various venues. We have come with three round approach:¶

Round 1:¶

✅ Test data (nearly 20%) will be sent to you for inference  --> NO NEED FOR MAKING DOCKER at this stage (we want to give everyone chance to do conveniently)✅ You will need to zip the masks and submit it to the leaderboard that will open on 17th February🚫 You are not allowed to use this for any training or do any manual annotation (if found the team team will be blacklisted and barred from all EndoCV challenge series)
⬆️ Elimination criteria: Method (novelty, use of sequence, unpublished)
🔖 Workshop proceeding: All teams are eligible for submitting a workshop paper (one paper per team only). The paper will be peer reviewed on CMT. More information will be provided closer to the dates. This year it will be published as IEEE ISBI2022 workshop proceedings as advised by the IEEE ISBI chairs  (agreement not reached!!!) CEUR proceeding  

Round 2:¶

✅ Test data (nearly 50%) will be sent to you for inference  --> NO NEED FOR MAKING DOCKER at this stage (we want to give everyone chance to do conveniently)✅ You will need to zip the masks and submit it to the leaderboard that will open on 20th February🚫 You are not allowed to use this for any training or do any manual annotation (if found the team team will be blacklisted and barred from all EndoCV challenge series)
⬆️ Elimination criteria: Scores must be higher than baseline score put by the organiser (selection of up to top 7 teams)

Round 3:¶

✅ Evaluation on all 100% test data (by organisers - we will also test fairness of your method reported in other two previous rounds, test inference time)✅ Models/inference code must be submitted as Docker image required *(we will help selected teams to achieve this!)✅ Evaluation internally and will be kept for journal publication
🛂 *EndoCV committee reserve the right to eliminate teams at any time if they find any violation of code-of-conduct of challenge and rules set in the challenge 
⬆️ Elimination criteria: Score on accuracy vs speed; robustness on unseen center; unseen modality; fairness; novelty (selection of top 3-5 teams per task will be invited for the joint-paper*)
*Selected teams will be eligible for joint journal paper participation only if they were successful in the peer reviewed workshop paper publication