Open Richar-Du opened 10 months ago
Thanks for your awesome work! I have a question about the order of the number mark: the mark allocation algorithm in the paper states that the areas are sorted in an ascending order so that smaller regions are considered before large regions. However, the code sorts them in a descending order: https://github.com/microsoft/SoM/blob/45ed34228cbf11457f0a131e6b78c7c8eb8e8d53/task_adapter/semantic_sam/tasks/inference_semsam_m2m_auto.py#L40C23-L40C23 And the largest area is marked first:
Do I misunderstand it?
Thanks for your awesome work! I have a question about the order of the number mark: the mark allocation algorithm in the paper states that the areas are sorted in an ascending order so that smaller regions are considered before large regions. However, the code sorts them in a descending order: https://github.com/microsoft/SoM/blob/45ed34228cbf11457f0a131e6b78c7c8eb8e8d53/task_adapter/semantic_sam/tasks/inference_semsam_m2m_auto.py#L40C23-L40C23 And the largest area is marked first:
Do I misunderstand it?