Participation Policies¶

Registration Form¶

In order to be eligible for final ranking and prizes, teams are required to read and fill in a registration form. Only 1 form per team is needed; a participant may only be a member of 1 team. A team can be composed of only 1 participant if desired. 

Link to PDF version of the registration form: https://drive.google.com/file/d/1TRzbsXJTxuww1IR3DjHRTOwJnFhVR7mt/view?usp=sharing 

Please download and return the filled form to hntsmrg2024@gmail.com with the following subject heading and be sure to include your team name — “HNTSMRG 2024: Registration Form – INSERT TEAM NAME HERE”. 

Important Information¶

• The Training dataset (images and segmentation masks) will be made publicly available to all participants through an online repository via a Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0). This means participants are free to utilize data given appropriate attribution for noncommercial purposes. 
• All data is considered de-identified information under the Health Insurance Portability and Accountability Act (HIPAA). Regardless, participants shall not use the data, either alone or in combination with other information, to attempt to identify or contact any individuals potentially represented in these data. Additionally, participants shall not use data to reconstruct any physical features of the face, head, or body. 
• The Testing dataset will not be made public during the challenge period. Participant algorithms will be assessed via Docker containers which will keep the test data hidden. 
• Participating teams will be granted a restricted number of submissions for evaluating their final algorithms using the hidden test set. Teams that are found to be deliberately employing multiple accounts to circumvent this limitation will be subject to disqualification at the discretion of the organizers.
• Teams will be required to submit a summary paper documenting their algorithm methods. See more details in the Paper Submission page. 
• Use of public data (e.g., for pre-training) is allowed. However, the use of private non-public data is not allowed. Teams who incorporate additional private external data sources into their algorithms will be exempt from any prize and final ranking eligibility, but are still encouraged to submit a summary paper for the challenge proceedings. 
• Teams are allowed to use pre-trained model weights/parameters (e.g., via transfer learning) if those models/weights are already public (e.g., ImageNet). Any use of pre-trained models must be disclosed in the submitted summary paper. 
• Members of the organizers' institutes (The University of Texas MD Anderson Cancer Center) may participate in the challenge but are not eligible for awards. 
• Following the conclusion of the challenge and a designated embargo period, the Training and Test datasets, along with their associated metadata, will be publicly released via an online repository.