cannot get the output segmentation in the paper with the provided weight file

[opnumten]

Hi,

with the provided weight file, I cannot get the output images shown in your PLOS paper. The prediction accuracy is pretty low.

Thanks

[vanvalen]

I'll take a look at it today! Dave David Van Valen MD, PhD Postdoctoral Scholar, Covert Lab Stanford University

…

On Mon, May 15, 2017 at 12:27 PM, opnumten ***@***.***> wrote: Hi, with the provided weight file, I cannot get the output images shown in your PLOS paper. The prediction accuracy is pretty low. Thanks — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#6>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AKYa9hWacS3smxsWR_-jiCUJ0rcU7E5dks5r6KckgaJpZM4NblEB> .

[xinrzhsh24]

Hi, Valen~
I can get the pretty segmentation results using your trained model ( '2016-07-12_nuclei_all_61x61_bn_feature_net_61x61_'). But the model trained by using your code of 'training_template', the result is too bad. I also used your data (nulei_all_6161) and the model bn_feature_net_61*61. I noticed your trained model is 1.9M, while mine is 3.8M. What is the problems?

[hftsai]

I'm seeing similar..i'm using just the trained network provided and the validation data provided (HeLa)
The Jaccard index and dice index provided in the original jupyter notebook suggest:
Jaccard index is 0.84462 +/- 0.0107597496164
Dice index is 0.914696973088 +/- 0.00655002244584

But when i run it on the docker i pulled
the indexes are only:
Jaccard index is 0.700539 +/- 0.144342
Dice index is 0.814186574197 +/- 0.115389352237b

does anyone know why this would happen?

[vanvalen]

I'm taking a look at it. Dave David Van Valen MD, PhD Postdoctoral Scholar, Covert Lab Stanford University

…

On Tue, Jun 20, 2017 at 12:19 AM, Hsieh-Fu Tsai ***@***.***> wrote: I'm seeing similar..i'm using just the trained network provided and the validation data provided (HeLa) The Jaccard index and dice index provided in the original jupyter notebook suggest: Jaccard index is 0.84462 +/- 0.0107597496164 Dice index is 0.914696973088 +/- 0.00655002244584 But when i run it on the docker i pulled the indexes are only: Jaccard index is 0.700539 +/- 0.144342 Dice index is 0.814186574197 +/- 0.115389352237b does anyone know why this would happen? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#6 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AKYa9va4Om5FmApZBPhwbQ8piI-NnU2Eks5sF3J7gaJpZM4NblEB> .

[hftsai]

Hi Dave, Thank you so much for your time! I'm sure it'll be extremely useful for my work if we can take a ride on your platform.

[vanvalen]

I'm not quite sure what the issue is, but the fastest fix is to train new neural networks for the HeLa cells. They should be ready in a day or so. David Van Valen MD, PhD Postdoctoral Scholar, Covert Lab Stanford University

…

On Wed, Jun 21, 2017 at 1:20 PM, Hsieh-Fu Tsai ***@***.***> wrote: Hi Dave, Thank you so much for your time! I'm sure it'll be extremely useful for my work if we can take a ride on your platform. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#6 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AKYa9iNihC7bHxF5T-c8l0tQHdbfZbNdks5sGXsIgaJpZM4NblEB> .