From a372c4643622d49f91e483f3b3d56d7122994e94 Mon Sep 17 00:00:00 2001
From: Christopher Rhodes <christopher.rhodes@embl.de>
Date: Wed, 20 Dec 2023 13:27:49 +0100
Subject: [PATCH] Test and example notebook that covers object classification
from binary segmentation
---
conf/testing.py | 2 +-
extensions/ilastik/models.py | 15 +++++++++++----
extensions/ilastik/tests/test_ilastik.py | 18 +++++++++++++++++-
model_server/workflows.py | 2 +-
4 files changed, 30 insertions(+), 7 deletions(-)
diff --git a/conf/testing.py b/conf/testing.py
index 33cc0859..9a08092f 100644
--- a/conf/testing.py
+++ b/conf/testing.py
@@ -56,7 +56,7 @@ monozstackmask = {
ilastik_classifiers = {
'px': root / 'ilastik' / 'demo_px.ilp',
'pxmap_to_obj': root / 'ilastik' / 'demo_obj.ilp',
- 'seg_to_obj': root / 'ilastik' / 'new_auto_obj.ilp',
+ 'seg_to_obj': root / 'ilastik' / 'demo_obj_seg.ilp',
}
output_path = root / 'testing_output'
diff --git a/extensions/ilastik/models.py b/extensions/ilastik/models.py
index 864413d8..f39a78f8 100644
--- a/extensions/ilastik/models.py
+++ b/extensions/ilastik/models.py
@@ -92,9 +92,16 @@ class IlastikObjectClassifierFromSegmentationModel(IlastikModel, InstanceSegment
return ObjectClassificationWorkflowBinary
def infer(self, input_img: GenericImageDataAccessor, segmentation_img: GenericImageDataAccessor) -> (np.ndarray, dict):
- assert segmentation_img.is_mask()
tagged_input_data = vigra.taggedView(input_img.data, 'yxcz')
- tagged_seg_data = vigra.taggedView(segmentation_img.data, 'yxcz')
+ assert segmentation_img.is_mask()
+ if segmentation_img.dtype == 'bool':
+ seg = 255 * segmentation_img.data.astype('uint8')
+ tagged_seg_data = vigra.taggedView(
+ 255 * segmentation_img.data.astype('uint8'),
+ 'yxcz'
+ )
+ else:
+ tagged_seg_data = vigra.taggedView(segmentation_img.data, 'yxcz')
dsi = [
{
@@ -105,7 +112,7 @@ class IlastikObjectClassifierFromSegmentationModel(IlastikModel, InstanceSegment
obmaps = self.shell.workflow.batchProcessingApplet.run_export(dsi, export_to_array=True) # [z x h x w x n]
- assert (len(obmaps) == 1, 'ilastik generated more than one object map')
+ assert len(obmaps) == 1, 'ilastik generated more than one object map'
yxcz = np.moveaxis(
obmaps[0],
@@ -141,7 +148,7 @@ class IlastikObjectClassifierFromPixelPredictionsModel(IlastikModel, ImageToImag
obmaps = self.shell.workflow.batchProcessingApplet.run_export(dsi, export_to_array=True) # [z x h x w x n]
- assert (len(obmaps) == 1, 'ilastik generated more than one object map')
+ assert len(obmaps) == 1, 'ilastik generated more than one object map'
yxcz = np.moveaxis(
obmaps[0],
diff --git a/extensions/ilastik/tests/test_ilastik.py b/extensions/ilastik/tests/test_ilastik.py
index ed74c43e..1dfbde16 100644
--- a/extensions/ilastik/tests/test_ilastik.py
+++ b/extensions/ilastik/tests/test_ilastik.py
@@ -82,7 +82,7 @@ class TestIlastikPixelClassification(unittest.TestCase):
self.mono_image = mono_image
self.mask = mask
- def test_run_object_classifier(self):
+ def test_run_object_classifier_from_pixel_predictions(self):
self.test_run_pixel_classifier()
fp = czifile['path']
model = ilm.IlastikObjectClassifierFromPixelPredictionsModel(
@@ -98,6 +98,22 @@ class TestIlastikPixelClassification(unittest.TestCase):
)
self.assertEqual(objmap.data.max(), 3)
+ def test_run_object_classifier_from_segmentation(self):
+ self.test_run_pixel_classifier()
+ fp = czifile['path']
+ model = ilm.IlastikObjectClassifierFromSegmentationModel(
+ {'project_file': ilastik_classifiers['seg_to_obj']}
+ )
+ objmap = model.label_instance_class(self.mono_image, self.mask)
+
+ self.assertTrue(
+ write_accessor_data_to_file(
+ output_path / f'obmap_from_seg_{fp.stem}.tif',
+ objmap,
+ )
+ )
+ self.assertEqual(objmap.data.max(), 3)
+
def test_ilastik_pixel_classification_as_workflow(self):
result = infer_image_to_image(
czifile['path'],
diff --git a/model_server/workflows.py b/model_server/workflows.py
index c81d2afd..6eb7b50d 100644
--- a/model_server/workflows.py
+++ b/model_server/workflows.py
@@ -43,7 +43,7 @@ def infer_image_to_image(fpi: Path, model: Model, where_output: Path, **kwargs)
img = generate_file_accessor(fpi).get_one_channel_data(ch)
ti.click('file_input')
- outdata, _ = model.infer(img)
+ outdata = model.label_pixel_class(img)
ti.click('inference')
outpath = where_output / (model.model_id + '_' + fpi.stem + '.tif')
--
GitLab