cellmil.features.extractor.embedding¶

Classes

`EmbeddingExtractor`(extractor_name)
`GigapathExtractor`()	Gigapath feature extractor using a pretrained EfficientNet model from timm.
`ResNet50Extractor`()	ResNet50 feature extractor with adaptive mean pooling after 3rd residual block.
`UNIExtractor`()	UNI feature extractor using a pretrained model from timm.

class cellmil.features.extractor.embedding.EmbeddingExtractor(extractor_name: ExtractorType)[source]¶

Bases: object

__init__(extractor_name: ExtractorType)[source]¶

extract_features(batch: Tensor) → Tensor[source]¶

class cellmil.features.extractor.embedding.ResNet50Extractor[source]¶

Bases: object

ResNet50 feature extractor with adaptive mean pooling after 3rd residual block.

__init__()[source]¶

extract_features(batch: Tensor) → Tensor[source]¶

class cellmil.features.extractor.embedding.GigapathExtractor[source]¶

Bases: object

Gigapath feature extractor using a pretrained EfficientNet model from timm.

__init__()[source]¶

extract_features(batch: Tensor) → Tensor[source]¶

class cellmil.features.extractor.embedding.UNIExtractor[source]¶

Bases: object

UNI feature extractor using a pretrained model from timm.

Note: Requires huggingface_hub login with access token before first use. Run: from huggingface_hub import login; login()

__init__()[source]¶

extract_features(batch: Tensor) → Tensor[source]¶