Rules

By registering for, participating in, or submitting to the ORena SAVE FOCUS Challenge, participants confirm that they have read, understood, and agree to comply with these rules, the official platform terms, and any additional policies published on this challenge website.

This page provides a summary of the challenge rules. More detailed information can be found in the policy and design documents linked in the resources section at the bottom of this page.

General Guidelines 📜

The ORena SAVE FOCUS Challenge is hosted within the ORena framework and evaluated through the official challenge platform.
Participation is open to individuals, academia and industry teams, subject to applicable sanctions restrictions.
Due to funding restrictions, teams from countries or institutions subject to applicable U.S. sanctions or funding restrictions may not be permitted to participate.
All submitted methods must be fully automated. Manual interaction during evaluation is not possible.
Organizers reserve the right to disqualify teams for rule violations, unfair competition, platform abuse, account misuse, prohibited data use, or attempts to compromise challenge integrity.
Organizers may update technical constraints, schedules, or rules on the official website. Participants should monitor the Forum and Announcements pages for updates.

User Accounts and Teams 👥

Participation is organized at the team level. A team may consist of one or more participants.
Submission limits apply per team.
Creating multiple accounts or teams to bypass submission limits may result in exclusion.

Allowed Training Resources

FRAME and SEGMENT Track

Challenge data
Public datasets
Public pre-trained models

All training resources must be disclosed. External resources must be publicly available by the start of the pre-evaluation phase. Using internally additional (manually) generated annotations on public data is allowed as long as they are released by the submission deadline.

PROCEDURE Track

Challenge data
Public datasets
Private datasets
Open or closed models

External resources are not required to have been public at challenge launch.

Pre-Evaluation Phase 📈

There will be a public leaderboard for the pre-evaluation phase once submission opens.
Up to 10 submissions are allowed per team. The number is subject to change, depending on participant numbers.
The validation set comprises 20 unseen videos that are representative of the test set.
Two baseline models will be provided by the Challenge organizers per track and clearly marked in the leaderboard.

Test Phase Qualification

In order to qualify for the final submission, teams must successfully submit during the pre-evaluation phase.
For FRAME & SEGMENT tracks, a successful submission must beat both corresponding baselines.
For the PROCEDURE track, teams must beat both baselines on at least one leaderboard, technical or clinical.

Publication and Recognition 📄

Teams that beat the baselines may be invited as co-authors.
Up to three co-authors per team, with exceptions possible.

Conflict of Interest ⚖️

Members of organizer labs are not prize-eligible.
Organizer baselines will appear on the leaderboard and will be labeled.

Data Usage and Privacy 🔒

For a more detailed overview, please consult the policy and design documents:

Data Protection	Terms of Use	Challenge Rules
FRAME Track	SEGMENT Track	PROCEDURE Track