Add multi-batch prediction for the segmentation and Polaris consumers #189
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| @@ -191,11 +195,11 @@ def _load_data(self, redis_hash, subdir, fname): | |||
| labels = [frames[i] for i in range(num_frames)] | |||
|
|
|||
| # Cast y to int to avoid issues during fourier transform/drift correction | |||
| y = np.array(labels, dtype='uint16') | |||
| y = np.expand_dims(np.array(labels, dtype='uint16'), axis=-1) | |||
There was a problem hiding this comment.
Is this essentially adding back a channel dimension?
There was a problem hiding this comment.
Yeah, the tracking model needs it to be 4D and the segmentation consumer is currently returning a 3D image (BXY)
Comment on lines
199
to
+202
| # TODO: Why is there an extra dimension? | ||
| # Not a problem in tests, only with application based results. | ||
| # Issue with batch dimension from outputs? | ||
| y = y[:, 0] if y.shape[1] == 1 else y | ||
| # y = y[:, 0] if y.shape[1] == 1 else y |
There was a problem hiding this comment.
Thoughts on deleting vs. leaving commented out?
There was a problem hiding this comment.
I'm confused about when the 1st dimension would be 1 so I was worried that we were neglecting an edge case that would make this line relevant, but I don't feel strongly, we could delete it.
There was a problem hiding this comment.
Let's leave it just in case we end up debugging the edge case in the future.
Co-authored-by: Morgan Schwartz <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR adds support for multi-batch images for the segmentation and Polaris consumers. The segmentation consumer receives a variable
dimension_orderfrom the frontend (see this PR). This variable has been added to the data generated by the Polaris consumer for segmentation jobs.We have also added more comprehensive testing of the segmentation consumer for input images of different dimension orders. This involved changes to the
DummyStorageobject to allow the dimensions of the images it generates to be specified.We have removed a line from the
get_imageutils function that expands the dimensions of loaded TIFF files by default, because this behavior complicated the handling of images with dimensions BXYC or CXYB. This change had little impact on the other functions that callget_image, because this channel was often squeezed out immediately after loading.