Question about submission and validation
Hi,
Thank you for creating a wonderful challenge. I have some questions about the clarification on the 2D task of the challenge:
-Docker Submission Testing Files
Could you please specify which test files should be used to evaluate the Docker submission? The currently referenced files appear to belong to the segmentation task rather than this task.
-Input/Output Format for Docker Submission
It would be greatly helpful if you could clarify the expected input and output formats for the Docker file (e.g., JSON, images, etc.), along with any required formatting guidelines. A brief example would be much appreciated. I am assuming that for each modality, you will merge all datasets into one folder called images[Tr,Val,Ts] and a json file that contains all questions for that modality?
-Evaluation Metrics and Methodology
While the evaluation metrics are outlined, the input/output specifications are unclear, making it difficult to understand how these metrics will be calculated and whether to be included in our Docker file or only the output is required. Could you provide further details on the evaluation process? It would be great if the challenge provides a folder for evaluation script.
-CodaBench submission
From what i see in CodeBench, there are development phase (which has already passed in 2024?) and a testing phase that can be only submitted once. Is that intended? Also, it seems on the CodaBench that there is information about 2024 FLARE and 2025 FLARE which causes lots of confusion for me.
Thank you for your time answering my questions,
Le.
We’ve just updated the challenge page and released the baseline code — it should help clarify several of these points.
Also, we recommend joining the upcoming webinar for more details on the tasks, data, and submission format. We believe many of your questions will be addressed there.