available_weights
available_weights()Registry names and what they are, for listing to a user.
Named to match available_weights() in the R package and available_transforms() here: everything a user can be offered a list of answers to the same verb.
Read the descriptions rather than only the names. What a model cannot do decides whether it answers your question: the nuclei weights are semantic, so touching nuclei come back as one region and a count taken from them would be wrong.
Returns
| Name | Type | Description |
|---|---|---|
| dict[str, str] | Registry name to description. A registry name carries a pinned commit, so it means | |
| dict[str, str] | one set of numbers today and the same set next year. |
listing = available_weights() sorted(listing) [‘bccd-yolo3’, ‘dsbowl-instances’, ‘dsbowl-unet’, ‘fives-unet’] “BBBC038v1” in listing[“dsbowl-unet”] True listing[“dsbowl-unet”].startswith(“U-Net, 256x256”) True
Two of those four are the same images read two ways, which is the case the descriptions have to carry rather than the names: one is semantic and one tells the objects apart.
“instance” in listing[“dsbowl-unet”].lower() True “One file, three scores” in listing[“dsbowl-instances”] True