dataeval.extractors¶

Feature extractors that transform input data into arrays.

All extractors implement the FeatureExtractor protocol (__call__(data) -> Array) and can be used standalone or passed to Embeddings for batching, caching, and memory-mapped storage.

Classes¶

`BoVWExtractor`	Computes Bag of Visual Words histograms using SIFT keypoints.
`ClassifierUncertaintyExtractor` Deprecated	Computes prediction entropy from a classification model for drift detection.
`ClasswiseUncertaintyExtractor`	Per-class prediction entropy distributions for detection models.
`DetectionGeometryExtractor`	Turn a detection Model’s normalized boxes into per-detection geometry rows.
`FlattenExtractor`	Simple NumPy-based feature extractor that flattens images to 1D vectors.
`OnnxExtractor`	Extracts embeddings via ONNX Runtime with lazy model loading.
`ScoresExtractor`	Wrap a MAITE classification/detection Model as a `FeatureExtractor`.
`TorchExtractor`	Extracts embeddings from a PyTorch model, with optional intermediate layer hooking.
`UncertaintyExtractor`	Per-instance prediction entropy as a drift feature.