James Dolezal
James Dolezal
**System information**. - Have I written custom code (as opposed to using a stock example script provided in Keras): No - OS Platform and Distribution (e.g., Linux Ubuntu 16.04): Issue...
I have an *.svs image made with an Aperio slide scanner, which apparently uses 33003 compression. I have no problem loading the image when reading directly from a file: ```...
## Feature Currently, the UQ implementation is built around the assumption that the target is a binary outcome, although some flexibility for multi-class uncertainty has already been implemented. This needs...
## Feature The Gaussian blur augmentation in the PyTorch backend differs slightly from the Tensorflow implementation, as it uses `torchvision.transforms.GaussianBlur`, requiring odd-numbered kernel sizes only. It is also much slower,...
## Feature Compatibility and feature testing when running Slideflow Studio from Docker containers (as referenced in #240) ## Additional context GLFW OpenGL rendering and Tkinter both theoretically support Docker, so...
# Notes: ## Loss For the torch models the loss for the time-to-event model was called [CoxProportionalHazardsLoss](https://github.com/jamesdolezal/slideflow/blob/f3cd55f287feb5f41ae9f25e933f20c2203c4f5e/slideflow/model/torch_utils.py#L514) this name is more specific than [negative_log_likelihood](https://github.com/jamesdolezal/slideflow/blob/f3cd55f287feb5f41ae9f25e933f20c2203c4f5e/slideflow/model/tensorflow_utils.py#L80) which is used for Tensorflow models....
**Summary** This new feature allows you to force each bag to draw the same amount of feature vectors from each slides. **Implementation** The bool `balanced` was added to :class:`TrainerConfigFastAI`. The...
# Add feature addition functionality to DatasetFeatures This PR adds a new method `add_features()` to the `DatasetFeatures` class, allowing users to incorporate additional features from annotation files into the existing...