Skip to content

model weights and config modification #2

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
kiryteo wants to merge 7 commits into
base: master
Choose a base branch
from
Open

model weights and config modification #2

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
8a5c6ff
model weights and config modification
kiryteo Dec 18, 2020
14f7b05
axes modification in the config
kiryteo Dec 18, 2020
f4a8e90
source distance idea
kiryteo Dec 20, 2020
bf1ada6
source displacement routine changes
kiryteo Dec 22, 2020
2d15769
Revert "source displacement routine changes"
kiryteo Dec 22, 2020
b5ddc38
new pairwise distance idea
kiryteo Dec 23, 2020
1bf5b55
check model script added
kiryteo Feb 8, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
18 changes: 9 additions & 9 deletions Bioimage-io-scripts/UNet3DKinetochores.model.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,10 +25,10 @@ timestamp: 2019-12-11T12:22:32Z # ISO 8601
inputs:
- name: raw
description: raw input
axes: bczyx # letters of axes in btczyx
axes: czyx # letters of axes in btczyx
data_type: float32
data_range: [-inf, inf]
shape: [1, 1, 48, 128, 128]
shape: [1, 48, 128, 128]
preprocessing: # list of preprocessing steps
- name: zero_mean_unit_variance # name of preprocessing step
kwargs:
Expand All @@ -48,11 +48,11 @@ outputs:
axes: bczyx
data_type: float32
data_range: [-inf, inf]
halo: [0, 0, 32, 48, 48]
halo: [0, 32, 48, 48]
Copy link
Copy Markdown
Member

@FynnBe FynnBe Dec 18, 2020

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

is really so much of the returned output affected by edge artefacts?

Copy link
Copy Markdown
Collaborator Author

@kiryteo kiryteo Dec 18, 2020

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is based on the settings in Pytorch-3DUNet, which worked well for Kinetochores use case (training as well as inference). But do you suggest to go for smaller values? (I can try that)

Copy link
Copy Markdown
Member

@FynnBe FynnBe Dec 18, 2020

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

could this be the halo within the network? You have your slicer, etc. to go over the whole volume. Just make sure that this is actually specifying the final output and not an intermediate step within your algorithm. If that's the case leave it as is and let's get this working before we start tweaking things.

Copy link
Copy Markdown
Collaborator Author

@kiryteo kiryteo Dec 18, 2020 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes, I checked for this and it is part of the predictor config and the routine.

Copy link
Copy Markdown
Member

@FynnBe FynnBe Dec 18, 2020

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

my understanding was that as this slicer is only used within your model it has no direct influence over the overall in- and output of the whole bioimage.io model. Let's take a closer look at this when we have a more or less running example.

shape:
reference_input: raw
scale: [1, 1, 1, 1, 1]
offset: [0, 0, 0, 0, 0]
scale: [1, 1, 1, 1]
offset: [0, 0, 0, 0]

language: python
framework: pytorch
Expand All @@ -63,11 +63,11 @@ dependencies: conda:../environment.yaml
test_inputs: [test_input.npy]
test_outputs: [test_output.npy]

sample_inputs: [sample_input.npy]
sample_outputs: [sample_output.npy]
# sample_inputs: [sample_input.npy]
# sample_outputs: [sample_output.npy]

weights:
pytorch_state_dict:
authors: [Ashwin Samudre;@bioimage-io]
sha256: e4d3885bccbe41cbf6c1d825f3cd2b707c7021ead5593156007e407a16b27cf2
source: https://zenodo.org/record/3446812/files/unet3d_kinetochores_weights.torch
#sha256: e4d3885bccbe41cbf6c1d825f3cd2b707c7021ead5593156007e407a16b27cf2
source: [best_checkpoint.pytorch]
Binary file added Bioimage-io-scripts/best_checkpoint.pytorch
View file
Open in desktop
Binary file not shown.