Dear organizers,
I would like to confirm the directory structures for the segmentation task.
According to https://github.com/schnadoslin/PhaKIR_Submission_Template?tab=readme-ov-file#testing-and-constraints, the input directory structure is;
inputs
│
├─── Video_01
│ ├─── Frames
│ │ ├─── frame_000000.png
│ │ ├─── frame_000001.png
│ │ └─── ...
And the output directory structure for the segmentation task is;
outputs
│
├───Video_01
│ └─── Video_01_Masks
│ ├─── 0
│ │ ├─── frame_000000.png
│ │ ├─── frame_000025.png
│ │ ├─── frame_000050.png
│ │ └─── ...
│ │
│ ├─── 1000
│ │ ├─── frame_001000.png
│ │ ├─── frame_001025.png
│ │ └─── ...
The structures are different in input and output. Is that correct ?
Dear Satoshi,
Thank you very much for your comment.
Yes, you are right, the structures are different for input and output, and the input structure of the test data is different from that of the training data, as the frames are not grouped in folders of 1000 but are specified directly. However, since the Github submission template you mentioned has been online for a long time and due to the short time until the deadline, we do not want to change the structure specified in the submission template, but would like to ask all participants to follow the structure given in the Github submission template repository.
Thank you very much for your thoughtful comment and best regards,
Tobias Rueckert