Dear sir, Should we make some models? For example, should we train Leave one "procedure and patient" out for stage 3 training? Thank you.

Created by takuya suzuki suzuki.t
Dear kkirtac, endoVis, Thank you very much for teaching. I understand.
Dear suzuki.t, kkirtac, please create only one model per task; don't make a difference between the stages - as kkirtac explained. Kind regards, the ROBUST-MIS organizers
We assume that we won't create a separate model for each stage, but rather create a single evaluation model per task. For instance, the model you submit for binary segmentation task will get used for evaluating stage 1, 2 and 3 of binary segmentation stages. I also request a clarification on this.

How many inference models should we create? page is loading…