<div align="center">
**Python library for 2D cell/nuclei instance segmentation models written with [PyTorch](https://pytorch.org/).**
[](https://github.com/okunator/cellseg_models.pytorch/blob/master/LICENSE)
[](https://pytorch.org/)
[](https://www.python.org/)
<br>
[](https://github.com/okunator/cellseg_models.pytorch/actions/workflows/tests.yml)
[](https://pypi.org/project/cellseg-models-pytorch/)
[](https://codecov.io/gh/okunator/cellseg_models.pytorch)
<br>
[](https://zenodo.org/badge/latestdoi/450787123)
</div>
<div align="center">
</div>
## Introduction
**cellseg-models.pytorch** is a library built upon [PyTorch](https://pytorch.org/) that contains multi-task encoder-decoder architectures along with dedicated post-processing methods for segmenting cell/nuclei instances. As the name might suggest, this library is heavily inspired by [segmentation_models.pytorch](https://github.com/qubvel/segmentation_models.pytorch) library for semantic segmentation.
## What's new? 📢
- Now you can use any pre-trained image encoder from the [timm](https://github.com/huggingface/pytorch-image-models) library as the model backbone. (Given that they implement the `forward_intermediates` method, most of them do).
- New example notebooks showing how to finetune **Cellpose** and **Stardist** with the new *state-of-the-art* foundation model backbones: [*UNI*](https://www.nature.com/articles/s41591-024-02857-3#Sec13) from the [MahmoodLab](https://faisal.ai/), and [Prov-GigaPath](https://www.nature.com/articles/s41586-024-07441-w) from Microsoft Research. Check out the notebooks [here (UNI)](https://github.com/okunator/cellseg_models.pytorch/blob/main/examples/pannuke_nuclei_segmentation_cellpose_UNI.ipynb), and [here (Prov-GigaPath)](https://github.com/okunator/cellseg_models.pytorch/blob/main/examples/pannuke_nuclei_segmentation_stardist_provgigapath.ipynb).
- **NOTE!**: **These foundation models are licensed under restrictive licences and you need to agree to the terms of any of the said models to get access to the weights. Once you have been granted access, you can run the above notebooks**. You can request for access here: [the model pages UNI](https://huggingface.co/MahmoodLab/UNI) and [model pages Prov-GigaPath](https://huggingface.co/prov-gigapath/prov-gigapath). These models may only be used for non-commercial, academic research purposes with proper attribution. **Be sure that you have read and understood the terms before requesting access**.
## Features 🌟
- High level API to define cell/nuclei instance segmentation models.
- 6 cell/nuclei instance segmentation model architectures and more to come.
- Open source datasets for training and benchmarking.
- Flexibility to modify the components of the model architectures.
- Sliding window inference for large images.
- Multi-GPU inference.
- All model architectures can be augmented to [panoptic segmentation](https://arxiv.org/abs/1801.00868).
- Popular training losses and benchmarking metrics.
- Benchmarking utilities both for model latency & segmentation performance.
- Regularization techniques to tackle batch effects/domain shifts such as [Strong Augment](https://arxiv.org/abs/2206.15274), [Spectral decoupling](https://arxiv.org/abs/2011.09468), [Label smoothing](https://arxiv.org/abs/1512.00567).
- Example notebooks to train models with [lightning](https://lightning.ai/docs/pytorch/latest/) or [accelerate](https://huggingface.co/docs/accelerate/index).
- Example notebooks to finetune models with foundation model backbones such as UNI, Prov-GigaPath, and DINOv2.
## Installation 🛠️
```shell
pip install cellseg-models-pytorch
```
## Models 🤖
| Model | Paper |
| -------------------------- | ------------------------------------------------------------------------------ |
| [[1](#Citation)] HoVer-Net | https://www.sciencedirect.com/science/article/pii/S1361841519301045?via%3Dihub |
| [[2](#Citation)] Cellpose | https://www.nature.com/articles/s41592-020-01018-x |
| [[3](#Citation)] Omnipose | https://www.biorxiv.org/content/10.1101/2021.11.03.467199v2 |
| [[4](#Citation)] Stardist | https://arxiv.org/abs/1806.03535 |
| [[5](#Citation)] CellVit-SAM | https://arxiv.org/abs/2306.15350.03535 |
| [[6](#Citation)] CPP-Net | https://arxiv.org/abs/2102.0686703535 |
## Datasets
| Dataset | Paper |
| ----------------------------- | ------------------------------------------------------------------------------------------------ |
| [[7, 8](#References)] Pannuke | https://arxiv.org/abs/2003.10778 , https://link.springer.com/chapter/10.1007/978-3-030-23937-4_2 |
## Notebook examples 👇
<details>
<summary style="margin-left: 25px;">Finetuning CellPose with UNI backbone</summary>
<div style="margin-left: 25px;">
- [Finetuning CellPose with UNI](https://github.com/okunator/cellseg_models.pytorch/blob/main/examples/pannuke_nuclei_segmentation_cellpose_UNI.ipynb). Here we finetune the CellPose multi-class nuclei segmentation model with the foundation model `UNI`-image-encoder backbone (checkout [UNI](https://www.nature.com/articles/s41591-024-02857-3#Sec13)). The Pannuke dataset (fold 1 & fold 2) are used for training data and the fold 3 is used as validation data. The model is trained (with checkpointing) by utilizing [accelerate](https://huggingface.co/docs/accelerate/index) by hugginface. **NOTE** that you need to have granted access to the UNI weights and agreed to the terms of the model to be able to run the notebook.
</div>
</details>
<details>
<summary style="margin-left: 25px;">Finetuning Stardist with Prov-GigaPath backbone</summary>
<div style="margin-left: 25px;">
- [Finetuning Stardist with Prov-GigaPath](https://github.com/okunator/cellseg_models.pytorch/blob/main/examples/pannuke_nuclei_segmentation_stardist_provgigapath.ipynb). Here we finetune the Stardist multi-class nuclei segmentation model with the foundation model `Prov-GigaPath`-image-encoder backbone (checkout [Prov-GigaPath](https://www.nature.com/articles/s41586-024-07441-w)). The Pannuke dataset (fold 1 & fold 2) are used for training data and the fold 3 is used as validation data. The model is trained (with checkpointing) by utilizing [accelerate](https://huggingface.co/docs/accelerate/index) by hugginface. **NOTE** that you need to have granted access to the Prov-GigaPath weights and agreed to the terms of the model to be able to run the notebook.
</div>
</details>
<details>
<summary style="margin-left: 25px;">Finetuning CellPose with DINOv2 backbone</summary>
<div style="margin-left: 25px;">
- [Finetuning CellPose with DINOv2 backbone for Pannuke](https://github.com/okunator/cellseg_models.pytorch/blob/main/examples/pannuke_nuclei_segmentation_cellpose_dinov2.ipynb). Here we finetune the CellPose multi-class nuclei segmentation model with a `LVD-142M` pretrained `DINOv2` backbone. The Pannuke dataset (fold 1 & fold 2) are used for training data and the fold 3 is used as validation data. The model is trained (with checkpointing) by utilizing [lightning](https://lightning.ai/docs/pytorch/latest/).
</div>
</details>
<details>
<summary style="margin-left: 25px;">Finetuning CellVit-SAM with Pannuke</summary>
<div style="margin-left: 25px;">
- [Finetuning CellVit-SAM with Pannuke](https://github.com/okunator/cellseg_models.pytorch/blob/main/examples/pannuke_nuclei_segmentation_cellvit.ipynb). Here we finetune the CellVit-SAM multi-class nuclei segmentation model with a `SA-1B` pretrained `SAM`-image-encoder backbone (checkout [SAM](https://github.com/facebookresearch/segment-anything)). The encoder is transformer based `VitDet`-model. The Pannuke dataset (fold 1 & fold 2) are used for training data and the fold 3 is used as validation data. The model is trained (with checkpointing) by utilizing [accelerate](https://huggingface.co/docs/accelerate/index) by hugginface.
</div>
</details>
<details>
<summary style="margin-left: 25px;"> Training Hover-Net with Pannuke</summary>
<div style="margin-left: 25px;">
- [Training Hover-Net with Pannuke](https://github.com/okunator/cellseg_models.pytorch/blob/main/examples/pannuke_nuclei_segmentation_hovernet.ipynb). Here we train the `Hover-Net` nuclei segmentation model with an `imagenet` pretrained `resnet50` backbone from the `timm` library. The Pannuke dataset (fold 1 & fold 2) are used for training data and the fold 3 is used as validation data. The model is trained by utilizing [lightning](https://lightning.ai/docs/pytorch/latest/).
</div>
</details>
<details>
<summary style="margin-left: 25px;">Training Stardist with Pannuke</summary>
<div style="margin-left: 25px;">
- [Training Stardist with Pannuke](https://github.com/okunator/cellseg_models.pytorch/blob/main/examples/pannuke_nuclei_segmentation_stardist.ipynb). Here we train the `Stardist` multi-class nuclei segmentation model with an `imagenet` pretrained `efficientnetv2_s` backbone from the `timm` library. The Pannuke dataset (fold 1 & fold 2) are used for training data and the fold 3 is used as validation data. The model is trained by utilizing [lightning](https://lightning.ai/docs/pytorch/latest/).
</div>
</details>
<details>
<summary style="margin-left: 25px;">Training CellPose with Pannuke</summary>
<div style="margin-left: 25px;">
- [Training CellPose with Pannuke](https://github.com/okunator/cellseg_models.pytorch/blob/main/examples/pannuke_nuclei_segmentation_cellpose.ipynb). Here we train the `CellPose` multi-class nuclei segmentation model with an `imagenet` pretrained `convnext_small` backbone from the `timm` library. The Pannuke dataset (fold 1 & fold 2) are used for training data and the fold 3 is used as validation data. The model is trained (with checkpointing) by utilizing [accelerate](https://huggingface.co/docs/accelerate/index) by hugginface.
</div>
</details>
<details>
<summary style="margin-left: 25px;">Training OmniPose with Pannuke</summary>
<div style="margin-left: 25px;">
- [Training OmniPose with Pannuke](https://github.com/okunator/cellseg_models.pytorch/blob/main/examples/pannuke_nuclei_segmentation_omnipose.ipynb). Here we train the OmniPose multi-class nuclei segmentation model with an `imagenet` pretrained `focalnet_small_lrf` backbone from the `timm` library. The Pannuke dataset (fold 1 & fold 2) are used for training data and the fold 3 is used as validation data. The model is trained (with checkpointing) by utilizing [accelerate](https://huggingface.co/docs/accelerate/index) by hugginface.
</div>
</details>
<details>
<summary style="margin-left: 25px;">Training CPP-Net with Pannuke</summary>
<div style="margin-left: 25px;">
- [Training CPP-Net with Pannuke](https://github.com/okunator/cellseg_models.pytorch/blob/main/examples/pannuke_nuclei_segmentation_cppnet.ipynb). Here we train the CPP-Net multi-class nuclei segmentation model with an `imagenet` pretrained `efficientnetv2_s` backbone from the `timm` library. The Pannuke dataset (fold 1 & fold 2) are used for training data and the fold 3 is used as validation data. The model is trained by utilizing [lightning](https://lightning.ai/docs/pytorch/latest/).
</div>
</details>
<details>
<summary style="margin-left: 25px;">Benchmarking Cellpose Trained on Pannuke</summary>
<div style="margin-left: 25px;">
- [Benchmarking Cellpose Trained on Pannuke](https://github.com/okunator/cellseg_models.pytorch/blob/main/examples/pannuke_cellpose_benchmark.ipynb). Here we run benchmarking for `Cellpose` that was trained on Pannuke. Both the model performance and latency benchmarking are covered.
</div>
</details>
## Code Examples 💻
**Define Cellpose for cell segmentation.**
```python
import cellseg_models_pytorch as csmp
import torch
model = csmp.models.cellpose_base(type_classes=5)
x = torch.rand([1, 3, 256, 256])
# NOTE: the outputs still need post-processing.
y = model(x) # {"cellpose": [1, 2, 256, 256], "type": [1, 5, 256, 256]}
```
**Define Cellpose for cell and tissue area segmentation (Panoptic segmentation).**
```python
import cellseg_models_pytorch as csmp
import torch
model = csmp.models.cellpose_plus(type_classes=5, sem_classes=3)
x = torch.rand([1, 3, 256, 256])
# NOTE: the outputs still need post-processing.
y = model(x) # {"cellpose": [1, 2, 256, 256], "type": [1, 5, 256, 256], "sem": [1, 3, 256, 256]}
```
**Define panoptic Cellpose model with more flexibility.**
```python
import cellseg_models_pytorch as csmp
# the model will include two decoder branches.
decoders = ("cellpose", "sem")
# and in total three segmentation heads emerging from the decoders.
heads = {
"cellpose": {"cellpose": 2, "type": 5},
"sem": {"sem": 3}
}
model = csmp.CellPoseUnet(
decoders=decoders, # cellpose and semantic decoders
heads=heads, # three output heads
depth=5, # encoder depth
out_channels=(256, 128, 64, 32, 16), # num out channels at each decoder stage
layer_depths=(4, 4, 4, 4, 4), # num of conv blocks at each decoder layer
style_channels=256, # num of style vector channels
enc_name="resnet50", # timm encoder
enc_pretrain=True, # imagenet pretrained encoder
long_skip="unetpp", # unet++ long skips ("unet", "unetpp", "unet3p")
merge_policy="sum", # concatenate long skips ("cat", "sum")
short_skip="residual", # residual short skips ("basic", "residual", "dense")
normalization="bcn", # batch-channel-normalization.
activation="gelu", # gelu activation.
convolution="wsconv", # weight standardized conv.
attention="se", # squeeze-and-excitation attention.
pre_activate=False, # normalize and activation after convolution.
)
x = torch.rand([1, 3, 256, 256])
# NOTE: the outputs still need post-processing.
y = model(x) # {"cellpose": [1, 2, 256, 256], "type": [1, 5, 256, 256], "sem": [1, 3, 256, 256]}
```
**Run HoVer-Net inference and post-processing with a sliding window approach.**
```python
import cellseg_models_pytorch as csmp
# define the model
model = csmp.models.hovernet_base(type_classes=5)
# define the final activations for each model output
out_activations = {"hovernet": "tanh", "type": "softmax", "inst": "softmax"}
# define whether to weight down the predictions at the image boundaries
# typically, models perform the poorest at the image boundaries and with
# overlapping patches this causes issues which can be overcome by down-
# weighting the prediction boundaries
out_boundary_weights = {"hovernet": True, "type": False, "inst": False}
# define the inferer
inferer = csmp.inference.SlidingWindowInferer(
model=model,
input_folder="/path/to/images/",
checkpoint_path="/path/to/model/weights/",
out_activations=out_activations,
out_boundary_weights=out_boundary_weights,
instance_postproc="hovernet", # THE POST-PROCESSING METHOD
normalization="percentile", # same normalization as in training
patch_size=(256, 256),
stride=128,
padding=80,
batch_size=8,
)
inferer.infer()
inferer.out_masks
# {"image1" :{"inst": [H, W], "type": [H, W]}, ..., "imageN" :{"inst": [H, W], "type": [H, W]}}
```
## Models API
Generally, the model building API enables the effortless creation of hard-parameter sharing multi-task encoder-decoder CNN architectures. The general architectural schema is illustrated in the below image.
<br><br>
### Class API
The class API enables the most flexibility in defining different model architectures. It borrows a lot from [segmentation_models.pytorch](https://github.com/qubvel/segmentation_models.pytorch) models API.
**Model classes**:
- `csmp.CellPoseUnet`
- `csmp.StarDistUnet`
- `csmp.HoverNet`
- `csmp.CellVitSAM`
**All of the models contain**:
- `model.encoder` - pretrained [timm](https://github.com/rwightman/pytorch-image-models) backbone for feature extraction.
- `model.{decoder_name}_decoder` - Models can have multiple decoders with unique names.
- `model.{head_name}_seg_head` - Model decoders can have multiple segmentation heads with unique names.
- `model.forward(x)` - forward pass.
- `model.forward_features(x)` - forward pass of the encoder and decoders. Returns enc and dec features
**Defining your own multi-task architecture**
For example, to define a multi-task architecture that has `resnet50` encoder, four decoders, and 5 output heads with `CellPoseUnet` architectural components, we could do this:
```python
import cellseg_models_pytorch as csmp
import torch
model = csmp.CellPoseUnet(
decoders=("cellpose", "dist", "contour", "sem"),
heads={
"cellpose": {"type": 5, "cellpose": 2},
"dist": {"dist": 1},
"contour": {"contour": 1},
"sem": {"sem": 4}
},
)
x = torch.rand([1, 3, 256, 256])
model(x)
# {
# "cellpose": [1, 2, 256, 256],
# "type": [1, 5, 256, 256],
# "dist": [1, 1, 256, 256],
# "contour": [1, 1, 256, 256],
# "sem": [1, 4, 256, 256]
# }
```
### Function API
With the function API, you can build models with low effort by calling the below listed functions. Under the hood, the function API simply calls the above classes with pre-defined decoder and head names. The training and post-processing tools of this library are built around these names, thus, it is recommended to use the function API, although, it is a bit more rigid than the class API. Basically, the function API only lacks the ability to define the output-tasks of the model, but allows for all the rest as the class API.
| Model functions | Output names | Task |
| -------------------------------------- | ----------------------------------------- | -------------------------------- |
| `csmp.models.cellpose_base` | `"type"`, `"cellpose"`, | **instance segmentation** |
| `csmp.models.cellpose_plus` | `"type"`, `"cellpose"`, `"sem"`, | **panoptic segmentation** |
| `csmp.models.omnipose_base` | `"type"`, `"omnipose"` | **instance segmentation** |
| `csmp.models.omnipose_plus` | `"type"`, `"omnipose"`, `"sem"`, | **panoptic segmentation** |
| `csmp.models.hovernet_base` | `"type"`, `"inst"`, `"hovernet"` | **instance segmentation** |
| `csmp.models.hovernet_plus` | `"type"`, `"inst"`, `"hovernet"`, `"sem"` | **panoptic segmentation** |
| `csmp.models.hovernet_small` | `"type"`,`"hovernet"` | **instance segmentation** |
| `csmp.models.hovernet_small_plus` | `"type"`, `"hovernet"`, `"sem"` | **panoptic segmentation** |
| `csmp.models.stardist_base` | `"stardist"`, `"dist"` | **binary instance segmentation** |
| `csmp.models.stardist_base_multiclass` | `"stardist"`, `"dist"`, `"type"` | **instance segmentation** |
| `csmp.models.stardist_plus` | `"stardist"`, `"dist"`, `"type"`, `"sem"` | **panoptic segmentation** |
| `csmp.models.cppnet_base` | `"stardist_refined"`, `"dist"` | **binary instance segmentation** |
| `csmp.models.cppnet_base_multiclass` | `"stardist_refined"`, `"dist"`, `"type"` | **instance segmentation** |
| `csmp.models.cppnet_plus` | `"stardist_refined"`, `"dist"`, `"type"`, `"sem"` | **panoptic segmentation** |
| `csmp.models.cellvit_sam_base` | `"type"`, `"inst"`, `"hovernet"` | **instance segmentation** |
| `csmp.models.cellvit_sam_plus` | `"type"`, `"inst"`, `"hovernet"`, `"sem"` | **panoptic segmentation** |
| `csmp.models.cellvit_sam_small` | `"type"`,`"hovernet"` | **instance segmentation** |
| `csmp.models.cellvit_sam_small_plus` | `"type"`, `"hovernet"`, `"sem"` | **panoptic segmentation** |
## References
- [1] S. Graham, Q. D. Vu, S. E. A. Raza, A. Azam, Y-W. Tsang, J. T. Kwak and N. Rajpoot. "HoVer-Net: Simultaneous Segmentation and Classification of Nuclei in Multi-Tissue Histology Images." Medical Image Analysis, Sept. 2019.
- [2] Stringer, C.; Wang, T.; Michaelos, M. & Pachitariu, M. Cellpose: a generalist algorithm for cellular segmentation Nature Methods, 2021, 18, 100-106
- [3] Cutler, K. J., Stringer, C., Wiggins, P. A., & Mougous, J. D. (2022). Omnipose: a high-precision morphology-independent solution for bacterial cell segmentation. bioRxiv. doi:10.1101/2021.11.03.467199
- [4] Uwe Schmidt, Martin Weigert, Coleman Broaddus, & Gene Myers (2018). Cell Detection with Star-Convex Polygons. In Medical Image Computing and Computer Assisted Intervention - MICCAI 2018 - 21st International Conference, Granada, Spain, September 16-20, 2018, Proceedings, Part II (pp. 265–273).
- [5] Hörst, F., Rempe, M., Heine, L., Seibold, C., Keyl, J., Baldini, G., Ugurel, S., Siveke, J., Grünwald, B., Egger, J., & Kleesiek, J. (2023). CellViT: Vision Transformers for Precise Cell Segmentation and Classification (Version 1). arXiv. https://doi.org/10.48550/ARXIV.2306.15350.
- [6] Chen, S., Ding, C., Liu, M., Cheng, J., & Tao, D. (2023). CPP-Net: Context-Aware Polygon Proposal Network for Nucleus Segmentation. In IEEE Transactions on Image Processing (Vol. 32, pp. 980–994). Institute of Electrical and Electronics Engineers (IEEE). https://doi.org/10.1109/tip.2023.3237013
- [7] Gamper, J., Koohbanani, N., Benet, K., Khuram, A., & Rajpoot, N. (2019) PanNuke: an open pan-cancer histology dataset for nuclei instance segmentation and classification. In European Congress on Digital Pathology (pp. 11-19).
- [8] Gamper, J., Koohbanani, N., Graham, S., Jahanifar, M., Khurram, S., Azam, A.,Hewitt, K., & Rajpoot, N. (2020). PanNuke Dataset Extension, Insights and Baselines. arXiv preprint arXiv:2003.10778.
- [9] Graham, S., Jahanifar, M., Azam, A., Nimir, M., Tsang, Y.W., Dodd, K., Hero, E., Sahota, H., Tank, A., Benes, K., & others (2021). Lizard: A Large-Scale Dataset for Colonic Nuclear Instance Segmentation and Classification. In Proceedings of the IEEE/CVF International Conference on Computer Vision (pp. 684-693).
## Citation
```bibtex
@misc{csmp2022,
title={{cellseg_models.pytorch}: Cell/Nuclei Segmentation Models and Benchmark.},
author={Oskari Lehtonen},
howpublished = {\url{https://github.com/okunator/cellseg_models.pytorch}},
doi = {10.5281/zenodo.7064617}
year={2022}
}
```
## Licence
This project is distributed under [MIT License](https://github.com/okunator/cellseg_models.pytorch/blob/main/LICENSE)
The project contains code from the original cell segmentation and 3rd-party libraries that have permissive licenses:
- [timm](https://github.com/rwightman/pytorch-image-models) (Apache-2)
- [HoVer-Net](https://github.com/vqdang/hover_net) (MIT)
- [Cellpose](https://github.com/MouseLand/cellpose) (BSD-3)
- [Stardist](https://github.com/stardist/stardist) (BSD-3)
If you find this library useful in your project, it is your responsibility to ensure you comply with the conditions of any dependent licenses. Please create an issue if you think something is missing regarding the licenses.
Raw data
{
"_id": null,
"home_page": "https://github.com/okunator/cellseg_models.pytorch",
"name": "cellseg_models_pytorch",
"maintainer": null,
"docs_url": null,
"requires_python": "<4.0,>=3.10",
"maintainer_email": null,
"keywords": "cell segmentation, nuclei segmentation, pytorch, pytorch-lightning",
"author": "Okunator",
"author_email": "oskari.lehtonen@helsinki.fi",
"download_url": "https://files.pythonhosted.org/packages/11/ae/deae1efd75f295cca8aa331f7393f7137f97d6e93a01868d91795ed1443f/cellseg_models_pytorch-0.1.25.tar.gz",
"platform": null,
"description": "<div align=\"center\">\n\n\n\n**Python library for 2D cell/nuclei instance segmentation models written with [PyTorch](https://pytorch.org/).**\n\n[](https://github.com/okunator/cellseg_models.pytorch/blob/master/LICENSE)\n[](https://pytorch.org/)\n[](https://www.python.org/)\n<br>\n[](https://github.com/okunator/cellseg_models.pytorch/actions/workflows/tests.yml)\n[](https://pypi.org/project/cellseg-models-pytorch/)\n[](https://codecov.io/gh/okunator/cellseg_models.pytorch)\n<br>\n[](https://zenodo.org/badge/latestdoi/450787123)\n\n</div>\n\n<div align=\"center\">\n\n</div>\n\n## Introduction\n\n**cellseg-models.pytorch** is a library built upon [PyTorch](https://pytorch.org/) that contains multi-task encoder-decoder architectures along with dedicated post-processing methods for segmenting cell/nuclei instances. As the name might suggest, this library is heavily inspired by [segmentation_models.pytorch](https://github.com/qubvel/segmentation_models.pytorch) library for semantic segmentation. \n\n## What's new? \ud83d\udce2\n- Now you can use any pre-trained image encoder from the [timm](https://github.com/huggingface/pytorch-image-models) library as the model backbone. (Given that they implement the `forward_intermediates` method, most of them do).\n- New example notebooks showing how to finetune **Cellpose** and **Stardist** with the new *state-of-the-art* foundation model backbones: [*UNI*](https://www.nature.com/articles/s41591-024-02857-3#Sec13) from the [MahmoodLab](https://faisal.ai/), and [Prov-GigaPath](https://www.nature.com/articles/s41586-024-07441-w) from Microsoft Research. Check out the notebooks [here (UNI)](https://github.com/okunator/cellseg_models.pytorch/blob/main/examples/pannuke_nuclei_segmentation_cellpose_UNI.ipynb), and [here (Prov-GigaPath)](https://github.com/okunator/cellseg_models.pytorch/blob/main/examples/pannuke_nuclei_segmentation_stardist_provgigapath.ipynb). \n- **NOTE!**: **These foundation models are licensed under restrictive licences and you need to agree to the terms of any of the said models to get access to the weights. Once you have been granted access, you can run the above notebooks**. You can request for access here: [the model pages UNI](https://huggingface.co/MahmoodLab/UNI) and [model pages Prov-GigaPath](https://huggingface.co/prov-gigapath/prov-gigapath). These models may only be used for non-commercial, academic research purposes with proper attribution. **Be sure that you have read and understood the terms before requesting access**. \n\n## Features \ud83c\udf1f\n\n- High level API to define cell/nuclei instance segmentation models.\n- 6 cell/nuclei instance segmentation model architectures and more to come.\n- Open source datasets for training and benchmarking.\n- Flexibility to modify the components of the model architectures.\n- Sliding window inference for large images.\n- Multi-GPU inference.\n- All model architectures can be augmented to [panoptic segmentation](https://arxiv.org/abs/1801.00868).\n- Popular training losses and benchmarking metrics.\n- Benchmarking utilities both for model latency & segmentation performance.\n- Regularization techniques to tackle batch effects/domain shifts such as [Strong Augment](https://arxiv.org/abs/2206.15274), [Spectral decoupling](https://arxiv.org/abs/2011.09468), [Label smoothing](https://arxiv.org/abs/1512.00567).\n- Example notebooks to train models with [lightning](https://lightning.ai/docs/pytorch/latest/) or [accelerate](https://huggingface.co/docs/accelerate/index).\n- Example notebooks to finetune models with foundation model backbones such as UNI, Prov-GigaPath, and DINOv2.\n\n\n## Installation \ud83d\udee0\ufe0f\n\n\n```shell\npip install cellseg-models-pytorch\n```\n\n## Models \ud83e\udd16\n\n| Model | Paper |\n| -------------------------- | ------------------------------------------------------------------------------ |\n| [[1](#Citation)] HoVer-Net | https://www.sciencedirect.com/science/article/pii/S1361841519301045?via%3Dihub |\n| [[2](#Citation)] Cellpose | https://www.nature.com/articles/s41592-020-01018-x |\n| [[3](#Citation)] Omnipose | https://www.biorxiv.org/content/10.1101/2021.11.03.467199v2 |\n| [[4](#Citation)] Stardist | https://arxiv.org/abs/1806.03535 |\n| [[5](#Citation)] CellVit-SAM | https://arxiv.org/abs/2306.15350.03535 |\n| [[6](#Citation)] CPP-Net | https://arxiv.org/abs/2102.0686703535 |\n\n## Datasets\n\n| Dataset | Paper |\n| ----------------------------- | ------------------------------------------------------------------------------------------------ |\n| [[7, 8](#References)] Pannuke | https://arxiv.org/abs/2003.10778 , https://link.springer.com/chapter/10.1007/978-3-030-23937-4_2 |\n\n## Notebook examples \ud83d\udc47\n\n<details>\n<summary style=\"margin-left: 25px;\">Finetuning CellPose with UNI backbone</summary>\n<div style=\"margin-left: 25px;\">\n\n- [Finetuning CellPose with UNI](https://github.com/okunator/cellseg_models.pytorch/blob/main/examples/pannuke_nuclei_segmentation_cellpose_UNI.ipynb). Here we finetune the CellPose multi-class nuclei segmentation model with the foundation model `UNI`-image-encoder backbone (checkout [UNI](https://www.nature.com/articles/s41591-024-02857-3#Sec13)). The Pannuke dataset (fold 1 & fold 2) are used for training data and the fold 3 is used as validation data. The model is trained (with checkpointing) by utilizing [accelerate](https://huggingface.co/docs/accelerate/index) by hugginface. **NOTE** that you need to have granted access to the UNI weights and agreed to the terms of the model to be able to run the notebook.\n\n</div>\n</details>\n\n<details>\n<summary style=\"margin-left: 25px;\">Finetuning Stardist with Prov-GigaPath backbone</summary>\n<div style=\"margin-left: 25px;\">\n\n- [Finetuning Stardist with Prov-GigaPath](https://github.com/okunator/cellseg_models.pytorch/blob/main/examples/pannuke_nuclei_segmentation_stardist_provgigapath.ipynb). Here we finetune the Stardist multi-class nuclei segmentation model with the foundation model `Prov-GigaPath`-image-encoder backbone (checkout [Prov-GigaPath](https://www.nature.com/articles/s41586-024-07441-w)). The Pannuke dataset (fold 1 & fold 2) are used for training data and the fold 3 is used as validation data. The model is trained (with checkpointing) by utilizing [accelerate](https://huggingface.co/docs/accelerate/index) by hugginface. **NOTE** that you need to have granted access to the Prov-GigaPath weights and agreed to the terms of the model to be able to run the notebook.\n\n</div>\n</details>\n\n<details>\n<summary style=\"margin-left: 25px;\">Finetuning CellPose with DINOv2 backbone</summary>\n<div style=\"margin-left: 25px;\">\n\n- [Finetuning CellPose with DINOv2 backbone for Pannuke](https://github.com/okunator/cellseg_models.pytorch/blob/main/examples/pannuke_nuclei_segmentation_cellpose_dinov2.ipynb). Here we finetune the CellPose multi-class nuclei segmentation model with a `LVD-142M` pretrained `DINOv2` backbone. The Pannuke dataset (fold 1 & fold 2) are used for training data and the fold 3 is used as validation data. The model is trained (with checkpointing) by utilizing [lightning](https://lightning.ai/docs/pytorch/latest/).\n\n</div>\n</details>\n\n<details>\n<summary style=\"margin-left: 25px;\">Finetuning CellVit-SAM with Pannuke</summary>\n<div style=\"margin-left: 25px;\">\n\n- [Finetuning CellVit-SAM with Pannuke](https://github.com/okunator/cellseg_models.pytorch/blob/main/examples/pannuke_nuclei_segmentation_cellvit.ipynb). Here we finetune the CellVit-SAM multi-class nuclei segmentation model with a `SA-1B` pretrained `SAM`-image-encoder backbone (checkout [SAM](https://github.com/facebookresearch/segment-anything)). The encoder is transformer based `VitDet`-model. The Pannuke dataset (fold 1 & fold 2) are used for training data and the fold 3 is used as validation data. The model is trained (with checkpointing) by utilizing [accelerate](https://huggingface.co/docs/accelerate/index) by hugginface.\n\n</div>\n</details>\n\n<details>\n<summary style=\"margin-left: 25px;\"> Training Hover-Net with Pannuke</summary>\n<div style=\"margin-left: 25px;\">\n\n- [Training Hover-Net with Pannuke](https://github.com/okunator/cellseg_models.pytorch/blob/main/examples/pannuke_nuclei_segmentation_hovernet.ipynb). Here we train the `Hover-Net` nuclei segmentation model with an `imagenet` pretrained `resnet50` backbone from the `timm` library. The Pannuke dataset (fold 1 & fold 2) are used for training data and the fold 3 is used as validation data. The model is trained by utilizing [lightning](https://lightning.ai/docs/pytorch/latest/).\n\n</div>\n</details>\n\n\n\n\n<details>\n<summary style=\"margin-left: 25px;\">Training Stardist with Pannuke</summary>\n<div style=\"margin-left: 25px;\">\n\n- [Training Stardist with Pannuke](https://github.com/okunator/cellseg_models.pytorch/blob/main/examples/pannuke_nuclei_segmentation_stardist.ipynb). Here we train the `Stardist` multi-class nuclei segmentation model with an `imagenet` pretrained `efficientnetv2_s` backbone from the `timm` library. The Pannuke dataset (fold 1 & fold 2) are used for training data and the fold 3 is used as validation data. The model is trained by utilizing [lightning](https://lightning.ai/docs/pytorch/latest/).\n\n</div>\n</details>\n\n<details>\n<summary style=\"margin-left: 25px;\">Training CellPose with Pannuke</summary>\n<div style=\"margin-left: 25px;\">\n\n- [Training CellPose with Pannuke](https://github.com/okunator/cellseg_models.pytorch/blob/main/examples/pannuke_nuclei_segmentation_cellpose.ipynb). Here we train the `CellPose` multi-class nuclei segmentation model with an `imagenet` pretrained `convnext_small` backbone from the `timm` library. The Pannuke dataset (fold 1 & fold 2) are used for training data and the fold 3 is used as validation data. The model is trained (with checkpointing) by utilizing [accelerate](https://huggingface.co/docs/accelerate/index) by hugginface.\n\n</div>\n</details>\n\n<details>\n<summary style=\"margin-left: 25px;\">Training OmniPose with Pannuke</summary>\n<div style=\"margin-left: 25px;\">\n\n- [Training OmniPose with Pannuke](https://github.com/okunator/cellseg_models.pytorch/blob/main/examples/pannuke_nuclei_segmentation_omnipose.ipynb). Here we train the OmniPose multi-class nuclei segmentation model with an `imagenet` pretrained `focalnet_small_lrf` backbone from the `timm` library. The Pannuke dataset (fold 1 & fold 2) are used for training data and the fold 3 is used as validation data. The model is trained (with checkpointing) by utilizing [accelerate](https://huggingface.co/docs/accelerate/index) by hugginface.\n\n</div>\n</details>\n\n<details>\n<summary style=\"margin-left: 25px;\">Training CPP-Net with Pannuke</summary>\n<div style=\"margin-left: 25px;\">\n\n- [Training CPP-Net with Pannuke](https://github.com/okunator/cellseg_models.pytorch/blob/main/examples/pannuke_nuclei_segmentation_cppnet.ipynb). Here we train the CPP-Net multi-class nuclei segmentation model with an `imagenet` pretrained `efficientnetv2_s` backbone from the `timm` library. The Pannuke dataset (fold 1 & fold 2) are used for training data and the fold 3 is used as validation data. The model is trained by utilizing [lightning](https://lightning.ai/docs/pytorch/latest/).\n\n</div>\n</details>\n\n<details>\n<summary style=\"margin-left: 25px;\">Benchmarking Cellpose Trained on Pannuke</summary>\n<div style=\"margin-left: 25px;\">\n\n- [Benchmarking Cellpose Trained on Pannuke](https://github.com/okunator/cellseg_models.pytorch/blob/main/examples/pannuke_cellpose_benchmark.ipynb). Here we run benchmarking for `Cellpose` that was trained on Pannuke. Both the model performance and latency benchmarking are covered.\n\n</div>\n</details>\n\n## Code Examples \ud83d\udcbb\n\n**Define Cellpose for cell segmentation.**\n\n```python\nimport cellseg_models_pytorch as csmp\nimport torch\n\nmodel = csmp.models.cellpose_base(type_classes=5)\nx = torch.rand([1, 3, 256, 256])\n\n# NOTE: the outputs still need post-processing.\ny = model(x) # {\"cellpose\": [1, 2, 256, 256], \"type\": [1, 5, 256, 256]}\n```\n\n**Define Cellpose for cell and tissue area segmentation (Panoptic segmentation).**\n\n```python\nimport cellseg_models_pytorch as csmp\nimport torch\n\nmodel = csmp.models.cellpose_plus(type_classes=5, sem_classes=3)\nx = torch.rand([1, 3, 256, 256])\n\n# NOTE: the outputs still need post-processing.\ny = model(x) # {\"cellpose\": [1, 2, 256, 256], \"type\": [1, 5, 256, 256], \"sem\": [1, 3, 256, 256]}\n```\n\n**Define panoptic Cellpose model with more flexibility.**\n\n```python\nimport cellseg_models_pytorch as csmp\n\n# the model will include two decoder branches.\ndecoders = (\"cellpose\", \"sem\")\n\n# and in total three segmentation heads emerging from the decoders.\nheads = {\n \"cellpose\": {\"cellpose\": 2, \"type\": 5},\n \"sem\": {\"sem\": 3}\n}\n\nmodel = csmp.CellPoseUnet(\n decoders=decoders, # cellpose and semantic decoders\n heads=heads, # three output heads\n depth=5, # encoder depth\n out_channels=(256, 128, 64, 32, 16), # num out channels at each decoder stage\n layer_depths=(4, 4, 4, 4, 4), # num of conv blocks at each decoder layer\n style_channels=256, # num of style vector channels\n enc_name=\"resnet50\", # timm encoder\n enc_pretrain=True, # imagenet pretrained encoder\n long_skip=\"unetpp\", # unet++ long skips (\"unet\", \"unetpp\", \"unet3p\")\n merge_policy=\"sum\", # concatenate long skips (\"cat\", \"sum\")\n short_skip=\"residual\", # residual short skips (\"basic\", \"residual\", \"dense\")\n normalization=\"bcn\", # batch-channel-normalization.\n activation=\"gelu\", # gelu activation.\n convolution=\"wsconv\", # weight standardized conv.\n attention=\"se\", # squeeze-and-excitation attention.\n pre_activate=False, # normalize and activation after convolution.\n)\n\nx = torch.rand([1, 3, 256, 256])\n\n# NOTE: the outputs still need post-processing.\ny = model(x) # {\"cellpose\": [1, 2, 256, 256], \"type\": [1, 5, 256, 256], \"sem\": [1, 3, 256, 256]}\n```\n\n**Run HoVer-Net inference and post-processing with a sliding window approach.**\n\n```python\nimport cellseg_models_pytorch as csmp\n\n# define the model\nmodel = csmp.models.hovernet_base(type_classes=5)\n\n# define the final activations for each model output\nout_activations = {\"hovernet\": \"tanh\", \"type\": \"softmax\", \"inst\": \"softmax\"}\n\n# define whether to weight down the predictions at the image boundaries\n# typically, models perform the poorest at the image boundaries and with\n# overlapping patches this causes issues which can be overcome by down-\n# weighting the prediction boundaries\nout_boundary_weights = {\"hovernet\": True, \"type\": False, \"inst\": False}\n\n# define the inferer\ninferer = csmp.inference.SlidingWindowInferer(\n model=model,\n input_folder=\"/path/to/images/\",\n checkpoint_path=\"/path/to/model/weights/\",\n out_activations=out_activations,\n out_boundary_weights=out_boundary_weights,\n instance_postproc=\"hovernet\", # THE POST-PROCESSING METHOD\n normalization=\"percentile\", # same normalization as in training\n patch_size=(256, 256),\n stride=128,\n padding=80,\n batch_size=8,\n)\n\ninferer.infer()\n\ninferer.out_masks\n# {\"image1\" :{\"inst\": [H, W], \"type\": [H, W]}, ..., \"imageN\" :{\"inst\": [H, W], \"type\": [H, W]}}\n```\n\n## Models API\n\nGenerally, the model building API enables the effortless creation of hard-parameter sharing multi-task encoder-decoder CNN architectures. The general architectural schema is illustrated in the below image.\n\n<br><br>\n\n\n### Class API\n\nThe class API enables the most flexibility in defining different model architectures. It borrows a lot from [segmentation_models.pytorch](https://github.com/qubvel/segmentation_models.pytorch) models API.\n\n**Model classes**:\n\n- `csmp.CellPoseUnet`\n- `csmp.StarDistUnet`\n- `csmp.HoverNet`\n- `csmp.CellVitSAM`\n\n**All of the models contain**:\n\n- `model.encoder` - pretrained [timm](https://github.com/rwightman/pytorch-image-models) backbone for feature extraction.\n- `model.{decoder_name}_decoder` - Models can have multiple decoders with unique names.\n- `model.{head_name}_seg_head` - Model decoders can have multiple segmentation heads with unique names.\n- `model.forward(x)` - forward pass.\n- `model.forward_features(x)` - forward pass of the encoder and decoders. Returns enc and dec features\n\n**Defining your own multi-task architecture**\n\nFor example, to define a multi-task architecture that has `resnet50` encoder, four decoders, and 5 output heads with `CellPoseUnet` architectural components, we could do this:\n\n```python\nimport cellseg_models_pytorch as csmp\nimport torch\n\nmodel = csmp.CellPoseUnet(\n decoders=(\"cellpose\", \"dist\", \"contour\", \"sem\"),\n heads={\n \"cellpose\": {\"type\": 5, \"cellpose\": 2},\n \"dist\": {\"dist\": 1},\n \"contour\": {\"contour\": 1},\n \"sem\": {\"sem\": 4}\n },\n)\n\nx = torch.rand([1, 3, 256, 256])\nmodel(x)\n# {\n# \"cellpose\": [1, 2, 256, 256],\n# \"type\": [1, 5, 256, 256],\n# \"dist\": [1, 1, 256, 256],\n# \"contour\": [1, 1, 256, 256],\n# \"sem\": [1, 4, 256, 256]\n# }\n```\n\n### Function API\n\nWith the function API, you can build models with low effort by calling the below listed functions. Under the hood, the function API simply calls the above classes with pre-defined decoder and head names. The training and post-processing tools of this library are built around these names, thus, it is recommended to use the function API, although, it is a bit more rigid than the class API. Basically, the function API only lacks the ability to define the output-tasks of the model, but allows for all the rest as the class API.\n\n| Model functions | Output names | Task |\n| -------------------------------------- | ----------------------------------------- | -------------------------------- |\n| `csmp.models.cellpose_base` | `\"type\"`, `\"cellpose\"`, | **instance segmentation** |\n| `csmp.models.cellpose_plus` | `\"type\"`, `\"cellpose\"`, `\"sem\"`, | **panoptic segmentation** |\n| `csmp.models.omnipose_base` | `\"type\"`, `\"omnipose\"` | **instance segmentation** |\n| `csmp.models.omnipose_plus` | `\"type\"`, `\"omnipose\"`, `\"sem\"`, | **panoptic segmentation** |\n| `csmp.models.hovernet_base` | `\"type\"`, `\"inst\"`, `\"hovernet\"` | **instance segmentation** |\n| `csmp.models.hovernet_plus` | `\"type\"`, `\"inst\"`, `\"hovernet\"`, `\"sem\"` | **panoptic segmentation** |\n| `csmp.models.hovernet_small` | `\"type\"`,`\"hovernet\"` | **instance segmentation** |\n| `csmp.models.hovernet_small_plus` | `\"type\"`, `\"hovernet\"`, `\"sem\"` | **panoptic segmentation** |\n| `csmp.models.stardist_base` | `\"stardist\"`, `\"dist\"` | **binary instance segmentation** |\n| `csmp.models.stardist_base_multiclass` | `\"stardist\"`, `\"dist\"`, `\"type\"` | **instance segmentation** |\n| `csmp.models.stardist_plus` | `\"stardist\"`, `\"dist\"`, `\"type\"`, `\"sem\"` | **panoptic segmentation** |\n| `csmp.models.cppnet_base` | `\"stardist_refined\"`, `\"dist\"` | **binary instance segmentation** |\n| `csmp.models.cppnet_base_multiclass` | `\"stardist_refined\"`, `\"dist\"`, `\"type\"` | **instance segmentation** |\n| `csmp.models.cppnet_plus` | `\"stardist_refined\"`, `\"dist\"`, `\"type\"`, `\"sem\"` | **panoptic segmentation** |\n| `csmp.models.cellvit_sam_base` | `\"type\"`, `\"inst\"`, `\"hovernet\"` | **instance segmentation** |\n| `csmp.models.cellvit_sam_plus` | `\"type\"`, `\"inst\"`, `\"hovernet\"`, `\"sem\"` | **panoptic segmentation** |\n| `csmp.models.cellvit_sam_small` | `\"type\"`,`\"hovernet\"` | **instance segmentation** |\n| `csmp.models.cellvit_sam_small_plus` | `\"type\"`, `\"hovernet\"`, `\"sem\"` | **panoptic segmentation** |\n\n## References\n\n- [1] S. Graham, Q. D. Vu, S. E. A. Raza, A. Azam, Y-W. Tsang, J. T. Kwak and N. Rajpoot. \"HoVer-Net: Simultaneous Segmentation and Classification of Nuclei in Multi-Tissue Histology Images.\" Medical Image Analysis, Sept. 2019.\n- [2] Stringer, C.; Wang, T.; Michaelos, M. & Pachitariu, M. Cellpose: a generalist algorithm for cellular segmentation Nature Methods, 2021, 18, 100-106\n- [3] Cutler, K. J., Stringer, C., Wiggins, P. A., & Mougous, J. D. (2022). Omnipose: a high-precision morphology-independent solution for bacterial cell segmentation. bioRxiv. doi:10.1101/2021.11.03.467199\n- [4] Uwe Schmidt, Martin Weigert, Coleman Broaddus, & Gene Myers (2018). Cell Detection with Star-Convex Polygons. In Medical Image Computing and Computer Assisted Intervention - MICCAI 2018 - 21st International Conference, Granada, Spain, September 16-20, 2018, Proceedings, Part II (pp. 265\u2013273).\n- [5] H\u00f6rst, F., Rempe, M., Heine, L., Seibold, C., Keyl, J., Baldini, G., Ugurel, S., Siveke, J., Gr\u00fcnwald, B., Egger, J., & Kleesiek, J. (2023). CellViT: Vision Transformers for Precise Cell Segmentation and Classification (Version 1). arXiv. https://doi.org/10.48550/ARXIV.2306.15350.\n- [6] Chen, S., Ding, C., Liu, M., Cheng, J., & Tao, D. (2023). CPP-Net: Context-Aware Polygon Proposal Network for Nucleus Segmentation. In IEEE Transactions on Image Processing (Vol. 32, pp. 980\u2013994). Institute of Electrical and Electronics Engineers (IEEE). https://doi.org/10.1109/tip.2023.3237013\n- [7] Gamper, J., Koohbanani, N., Benet, K., Khuram, A., & Rajpoot, N. (2019) PanNuke: an open pan-cancer histology dataset for nuclei instance segmentation and classification. In European Congress on Digital Pathology (pp. 11-19).\n- [8] Gamper, J., Koohbanani, N., Graham, S., Jahanifar, M., Khurram, S., Azam, A.,Hewitt, K., & Rajpoot, N. (2020). PanNuke Dataset Extension, Insights and Baselines. arXiv preprint arXiv:2003.10778.\n- [9] Graham, S., Jahanifar, M., Azam, A., Nimir, M., Tsang, Y.W., Dodd, K., Hero, E., Sahota, H., Tank, A., Benes, K., & others (2021). Lizard: A Large-Scale Dataset for Colonic Nuclear Instance Segmentation and Classification. In Proceedings of the IEEE/CVF International Conference on Computer Vision (pp. 684-693).\n\n## Citation\n\n```bibtex\n@misc{csmp2022,\n title={{cellseg_models.pytorch}: Cell/Nuclei Segmentation Models and Benchmark.},\n author={Oskari Lehtonen},\n howpublished = {\\url{https://github.com/okunator/cellseg_models.pytorch}},\n doi = {10.5281/zenodo.7064617}\n year={2022}\n}\n```\n\n## Licence\n\nThis project is distributed under [MIT License](https://github.com/okunator/cellseg_models.pytorch/blob/main/LICENSE)\n\nThe project contains code from the original cell segmentation and 3rd-party libraries that have permissive licenses:\n\n- [timm](https://github.com/rwightman/pytorch-image-models) (Apache-2)\n- [HoVer-Net](https://github.com/vqdang/hover_net) (MIT)\n- [Cellpose](https://github.com/MouseLand/cellpose) (BSD-3)\n- [Stardist](https://github.com/stardist/stardist) (BSD-3)\n\nIf you find this library useful in your project, it is your responsibility to ensure you comply with the conditions of any dependent licenses. Please create an issue if you think something is missing regarding the licenses.\n",
"bugtrack_url": null,
"license": "MIT",
"summary": "Python library for 2D cell/nuclei instance segmentation models written with PyTorch.",
"version": "0.1.25",
"project_urls": {
"Homepage": "https://github.com/okunator/cellseg_models.pytorch",
"Repository": "https://github.com/okunator/cellseg_models.pytorch"
},
"split_keywords": [
"cell segmentation",
" nuclei segmentation",
" pytorch",
" pytorch-lightning"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "7afd44861b4bbb836d1401d7940b063ecfbf0afe76ab28526d966276a13d8db0",
"md5": "4698d7f1480b0a0da50085872a018b27",
"sha256": "15344adba4401b99133dcabbfac53a3e34239d3250908ecd4b486c1091686d9d"
},
"downloads": -1,
"filename": "cellseg_models_pytorch-0.1.25-py3-none-any.whl",
"has_sig": false,
"md5_digest": "4698d7f1480b0a0da50085872a018b27",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": "<4.0,>=3.10",
"size": 4033372,
"upload_time": "2024-07-05T13:46:37",
"upload_time_iso_8601": "2024-07-05T13:46:37.080124Z",
"url": "https://files.pythonhosted.org/packages/7a/fd/44861b4bbb836d1401d7940b063ecfbf0afe76ab28526d966276a13d8db0/cellseg_models_pytorch-0.1.25-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "11aedeae1efd75f295cca8aa331f7393f7137f97d6e93a01868d91795ed1443f",
"md5": "4b0b5b960dbfa99292384b8c419edbf2",
"sha256": "12f29705e473f26a52eccbe55d422f5d7a83f44286b772a9a53dbbed9be05d5a"
},
"downloads": -1,
"filename": "cellseg_models_pytorch-0.1.25.tar.gz",
"has_sig": false,
"md5_digest": "4b0b5b960dbfa99292384b8c419edbf2",
"packagetype": "sdist",
"python_version": "source",
"requires_python": "<4.0,>=3.10",
"size": 3921776,
"upload_time": "2024-07-05T13:46:40",
"upload_time_iso_8601": "2024-07-05T13:46:40.112380Z",
"url": "https://files.pythonhosted.org/packages/11/ae/deae1efd75f295cca8aa331f7393f7137f97d6e93a01868d91795ed1443f/cellseg_models_pytorch-0.1.25.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-07-05 13:46:40",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "okunator",
"github_project": "cellseg_models.pytorch",
"travis_ci": false,
"coveralls": false,
"github_actions": true,
"lcname": "cellseg_models_pytorch"
}
JSON
