ikimhoerst/cellvit repository overview

⁠CellViT++

We created a beta version of a Docker image for CellViT++. This Docker container has all the necessary requirements preinstalled and is intended to run on a linux/amd64 platform. The container includes the CellViT-SAM-H checkpoint and all classifiers for this model.

• Paper: https://doi.org/10.48550/arXiv.2501.05269⁠
• Repo: https://github.com/TIO-IKIM/CellViT-plus-plus⁠

⁠Features

• Preinstalled Dependencies: The container comes with all required dependencies preinstalled, including PyTorch, CUDA, CuPy, NumPy, OpenSlide, Pillow, and more.
• GPU Support: The container is configured to utilize NVIDIA GPUs for accelerated computation.
• Preloaded Models: The CellViT-SAM-H checkpoint and classifiers are included within the container, eliminating the need for manual setup.

⁠Usage

To pull the image, use the following command:

docker pull ikimhoerst/cellvit:beta

⁠Running the Container

To start the container, use the following command. Replace /path/to/local/input and /path/to/local/output with the appropriate local paths on your system to load slides into the container as well as saving results locally:

docker run --name cellvit++ \
  --gpus all \ # gpus to use (default all)
  --memory=8g \ # memory, we recommend more (16g)
  -v /path/to/local/input:/workspace/CellViT-plus-plus/input-data \ This directory contains the input data mounted into the container.
  -v /path/to/local/output:/workspace/CellViT-plus-plus/output-data \ This directory contains the output data mounted from the container.
  -it \ # start terminal in workdir
  ikimhoerst/cellvit:beta

⁠Inference

To run inference, you can use the same commands as given in the examples given here (https://github.com/TIO-IKIM/CellViT-plus-plus?tab=readme-ov-file#examples⁠). However, you do not need to hand over the model checkpoint anymore, as CellViT-SAM-H is the default model in the container. Please consider your input and output paths accordingly. The classifiers are located inside the ./checkpoints/classifier/sam-h folder, similar to this git repository.

⁠License:

“Commons Clause” License Condition v1.0

The Software is provided to you by the Licensor under the License, as defined below, subject to the following condition.

Without limiting other conditions in the License, the grant of rights under the License will not include, and the License does not grant to you, the right to Sell the Software.

For purposes of the foregoing, “Sell” means practicing any or all of the rights granted to you under the License to provide to third parties, for a fee or other consideration (including without limitation fees for hosting or consulting/ support services related to the Software), a product or service whose value derives, entirely or substantially, from the functionality of the Software. Any license notice or attribution required by the License must also include this Commons Clause License Condition notice.

If any parts of this work is used, a citation of the following works is necessary: Hörst, F., Rempe, M., Heine, L., Seibold, C., Keyl, J., Baldini, G., Ugurel, S., Siveke, J., Grünwald, B., Egger, J., & Kleesiek, J. (2024). CellViT: Vision Transformers for precise cell segmentation and classification. In Medical Image Analysis (Vol. 94, p. 103143). Elsevier BV. https://doi.org/10.1016/j.media.2024.103143⁠

Hörst, F., Rempe, M., Becker, H., Heine, L., Keyl, J., & Kleesiek, J. (2025). CellViT++: Energy-Efficient and Adaptive Cell Segmentation and Classification Using Foundation Models (Version 1). arXiv. https://doi.org/10.48550/ARXIV.2501.05269⁠

Software: CellViT++

License: Apache 2.0 with Commons Clause

Licensor: Fabian Hörst, Jens Kleesiek