eva
eva copied to clipboard
Evaluation framework for oncology foundation models (FMs)
Closes #448 It saves the config file in the output folder and in the loggers: and optionally can be viewed in the terminal:
Currently we get the split from the dataloader_idx. This requires a `dataloader_idx_map` and a list of predict-datasets in the corresponding order. To make this more robust and allow one predict-dataset...
Currently `EmbeddingsWriter` can be used only for classification tasks. However, we would like to make it as a base class, and then sub-classes like `ClassificationEmbeddingsWriter` or `SegmentationEmbeddingsWriter` to be for...
This issue lists Renovate updates and detected dependencies. Read the [Dependency Dashboard](https://docs.renovatebot.com/key-concepts/dashboard/) docs to learn more.[View this repository on the Mend.io Web Portal](https://developer.mend.io/github/kaiko-ai/eva). ## Open The following updates have all...
We should clip the values of ct-scans from -1024 to 1024 and maybe applied windowing transform See https://stackoverflow.com/a/72562081
Main issue for all tasks concerning slide-level inference support. - [x] https://github.com/kaiko-ai/eva/issues/363 - [x] https://github.com/kaiko-ai/eva/issues/364 - [x] https://github.com/kaiko-ai/eva/issues/374 - [ ] https://github.com/kaiko-ai/eva/issues/375 - [x] https://github.com/kaiko-ai/eva/issues/376
Masking technique should be configurable. E.g. - precalculated foreground masks - on the fly tissue mask generation (e.g. using simple cv2 logic) - ...
For Nuclei Segmentation on CoNSeP dataset evaluation