Could you supply more samples in detail of performing your metric libraries in the format of realistic images, masks, and annotations?

[MIMIWAWA]

Thanks for your marvelous work!

I am writing my paper and preparing to quote your paper and use your metric library.

However, I am confused about the example on your Github.

Since the example of using "np.random.randint" is provided, could you supply more samples in detail of performing your metric libraries in the format of realistic images, masks, and annotations?

I appreciated having your examples in detail.

Thank you very much.

[muellerdo]

Heyho @MIMIWAWA,

thank you for your kind words! Always happy to hear that miseval is useful in other's work! :)

Sure, I added a more realistic detailed example (based on my paper Towards a Guideline for Evaluation Metrics in Medical Image Segmentation).

Check out this: https://github.com/frankkramer-lab/miseval/blob/master/miseval.example.ipynb

Overall, you have to convert your segmentation mask to a NumPy matrix in order to pass it to the miseval evaluate interface.

Hope that the new example makes things clearer :)

Cheers,
Dominik

[MIMIWAWA]

Dear @muellerdo,

Thank you so much for your detailed tutorial on "miseval.example.ipynb".
I have cited your paper, and I will cite it in my future published papers as well.

Moreover, if convenient for you, could you provide your codes for converting the mask file (.png) to a NumPy matrix (.npy) to keep the same track in the format?

I appreciated your help so much.

Sincerely yours,
MIMIWAWA