datamint

Plugin datamint from marketplace datamint in SonanceAI/datamint-python-api. Commands read from the repo's manifest — 2 days ago.

Skills (7)

• datamint-custom-models

  Turn a user's own model into a Datamint model with the `datamint` SDK - wrap any trained model (PyTorch checkpoint, state_dict, Hugging Face, timm, foundation models, models from a custom training loop) in a `DatamintModel` adapter, test it locally, validate it, and log/register it in the Datamint M

• datamint-data-access

  Explore, filter, load and download data from Datamint with the `datamint` Python SDK or the `datamint download` CLI. Covers listing projects, filtering resources (by project, tags, channel, modality, status, filename, date), loading images/DICOM/NIfTI/video into Python, slices and frames of volumes

• datamint-deploy

  Deploy registered Datamint models and run inference with the `datamint` SDK - starting and monitoring deployment jobs, choosing the model version, GPU deployments, managing deployed images, submitting remote inference jobs on resources (images, frames, slices, volumes, promptable models), saving pre

• datamint-end-to-end

  Plan and run a complete Datamint pipeline with the `datamint` SDK - from raw files to a deployed model - get data in, check annotations, build the dataset and splits, train, evaluate, deploy, run inference and send predictions back for review. Gives the order of stages, what each stage hands to the

• datamint-experiments

  Track, compare and manage Datamint experiments and models with MLflow and the `datamint` SDK - pointing MLflow at the Datamint server, logging params/metrics/models from custom training code, Lightning checkpoint callbacks that register models, finding and comparing runs, the model registry (`api.mo

• datamint-training

  Train Datamint models on Datamint projects with the `datamint` SDK - the built-in one-line trainers (UNet++, DeepLabV3+, TransUNet, UNETR++, nnU-Net, EfficientNetV2, timm classifiers, YOLOX), a custom architecture as a `SegmentationModule` / `ClassificationModule` subclass, PyTorch datasets (`ImageD

• datamint-upload

  Upload data to Datamint with the `datamint` Python SDK or the `datamint upload` / `datamint import` CLI. Covers DICOM files and series, NIfTI volumes, images and videos, with projects, tags, channels, metadata, anonymization, segmentation masks and AI pre-annotations for review in a worklist. Also c

Similar plugins