Label names are inconsistently transformed to lowercase

[leandervaneekelen]

Label names in the Label class are transformed to lowercase by default (good design choice imo):

pathology-whole-slide-data/wholeslidedata/annotation/labels.py

Line 26 in 1e4c6ca

self._name = str(name).lower()

However, there are multiple sampler classes that cross check labels by doing string comparisons, without doing a .lower() operation first, for example:

pathology-whole-slide-data/wholeslidedata/samplers/patchlabelsampler.py

Line 203 in 1e4c6ca

if annotation.label.name not in self._detection_labels:

This means that this behavior fails silently without any reason:

# Sample at some random coordinate where you expect there to be annotations of Foo
wsa = WholeSlideAnnotation('/path/to/some/file.xml', labels={'Foo': 1})
sampler = DetectionPatchLabelSampler(detection_labels=['Foo'])
annotations = sampler.sample(wsa, (10, 10), (512, 512), 1.0)  

>> annotations = []  # But it's empty because the labels are actually called 'foo' :(

Shall we fix this by just doing .lower by default on labels within all contexts?