The pre-trained weights are not available now

[caodoanh2001]

Hi, thank you for the good project.
However, two pre-trained models on pannuke and monusac datasets are not available anymore. Can you check again?

https://github.com/BiomedSciAI/histocartography/blob/dcad24866331bce4fb3d79b0565f23aec0001511/histocartography/preprocessing/nuclei_extraction.py#L31C1-L34C2

Thank you very much.

[Adali35]

I am getting the same problem, since I do not know any pretrained model other than the specified ones in the code I cannot run the nuclei_extraction. What should we do?

[yasohasakii]

Same issue with you, the pt file not in ibmbox now.

[yasohasakii]

Same issue with you, the pt file not in ibmbox now.

[caodoanh2001]

Nuclei extraction has the stage of nuclei segmentation using pretrained HoVerNet. I think we should use other inference code to extract the nuclei map instead of this repo. There were a couple of other implementations of HoVerNet for nuclei segmentation.

[Adali35]

I cannot find an available pretrained model for nuclei detection (in the .pt format), are there anyone that can help on this topic?

[ChrisBtt]

Did someone found a suitable pretrained model in order to execute the Nuclei detection or can an updated checkpoint file be provided?
I am not able to run the example code from the README at all without it.

[Adali35]

I actually got the hovernet model trained on pannuke dataset from original hovernet GitHub page and saved it locally using PyTorch. However this time when I used it in nucleidetection I got an error which says something like the "orderedDict does not have cpu() attribute" I didn't have time to resolve it honestly but are there someone who deals with similar problem?

[PushpakPati]

Sorry for the late respone.

The box links for HoVerNet checkpoints (monusac and pannuke) are broken now.
Until an update to the library, you can download them from Google drive at hovernet_monusac.pt and hovernet_pannuke

Please download the models and load them directly in the source code here.

[ChrisBtt]

Thank you very much for your fast solution. I am now able to run the introductory code and visualize the image including the graph network.