Skip to main content Crate klaster Copy item path Source metric Metrics used for evaluating clustering quality.
Provides Accuracy, ARI, and NMI. Autoencoder Convolutional autoencoder model used to learn latent embeddings.
Encodes images into a latent vector and reconstructs the input with a decoder. AutoencoderConfig Configuration for the Autoencoder model.
Defines convolutional and normalization parameters for the encoder/decoder stack. ClusteringOutput Holds embeddings, centroids, loss, and targets for metric computation and logging. Dataset Dataset for training and testing. Wraps raw image bytes and labels into train/test splits
and exposes helpers to build batches for SDC training. DatasetSplit Container for raw data. KMeans K-Means clustering model. KMeansFitted A fitted K-Means model containing learned cluster centroids and prediction methods. SDC SDC model implementation combining an autoencoder and clustering head. SDCConfig Configuration for the SDC model. TrainingConfig Configuration for training of an crate::SDC KMeansInit Initialization methods for KMeans clustering. infer Perform inference with a trained SDC model. Loads a saved model,
runs clustering on provided items, aligns clusters to labels and prints predictions to stdout. train Train the SDC model.