Registration¶

Anonymous registrations are not allowed. Participants must provide all required information during registration, ensuring it is complete and valid. Only participants with verified accounts are eligible to make submissions.

Use of Public Data and Models¶

Additional publicly available data and pre-trained models are allowed for training and validation. If used, participants should specify which pre-trained data and/or models were used, along with a public link for downloading them.

Submission Requirements¶

During the test phase, participants must submit both their algorithm and a report, including a GitHub repository URL. These items are not required during the validation phase. Refer to the Submission Instructions page for detailed guidelines.

Participation Rules¶

• Members of organizing centers may not participate in the challenge.
• All Docker containers submitted to the challenge will be run in an offline setting (i.e., they will not have access to the internet and cannot download/upload any resources). All necessary resources (e.g., pre-trained weights) must be encapsulated in the submitted containers beforehand.
• Participants and their AI algorithms must adhere to the compute limits. The participants are responsible for ensuring their algorithm runtime fits within the compute limits.
• The organizers of the PUMA challenge reserve the right to disqualify any participant or participating team at any time for unfair or dishonest practices.
• All participants reserve the right to withdraw from the challenge and forego further participation. However, they will not be able to retract their prior submissions or published results until then.

Training Data, Models, and Code¶

External data and pre-trained models are allowed in this competition as long as they are freely and publicly available under a permissive open-source license. Participants must clearly state the use of external data in their submission, using the algorithm name (e.g., "AI Model (trained w/ private data)"), algorithm page, and/or a supporting publication/URL.

The participating team’s code and model weights must be available on GitHub (or a similar platform) with a permissive open-source license.

Publication¶

• Up to three members of each leaderboard's top three performing teams will be invited to participate in the challenge paper as consortium authors.
• Participants in the PUMA challenge and non-participating researchers using the dataset can publish their results separately at any time.
• All submitted algorithms will be publicly available as Grand Challenge Algorithms.
• Any publication using this dataset must cite the PUMA challenge paper.

Additional Notes¶

The challenge organizers reserve the right to update the rules without prior notice.