Error during Line-Segmentation || TypeError: '<' not supported between instances of 'Image' and 'float'

[lsubhd]

Dear all,

once again I have come up with a problem during line segmentation:

File "/usr/local/lib/python3.6/dist-packages/ocr4all_helpers-0.2.2-py3.6.egg/ocr4all_helpers/pagelineseg.py", line 634, in cli
pool.map(parallel, dataset)
File "/usr/lib/python3.6/multiprocessing/pool.py", line 266, in map
return self._map_async(func, iterable, mapstar, chunksize).get()
File "/usr/lib/python3.6/multiprocessing/pool.py", line 644, in get
raise self._value
File "/usr/lib/python3.6/multiprocessing/pool.py", line 119, in worker
result = (True, func(*args, **kwds))
File "/usr/lib/python3.6/multiprocessing/pool.py", line 44, in mapstar
return list(map(*args))
File "/usr/local/lib/python3.6/dist-packages/ocr4all_helpers-0.2.2-py3.6.egg/ocr4all_helpers/pagelineseg.py", line 625, in parallel
remove_images=args.remove_images)
File "/usr/local/lib/python3.6/dist-packages/ocr4all_helpers-0.2.2-py3.6.egg/ocr4all_helpers/pagelineseg.py", line 369, in pagexmllineseg
cropped = Image.fromarray(nlbin.adaptive_binarize(np.array(cropped)).astype(np.uint8))
File "/usr/local/lib/python3.6/dist-packages/ocr4all_helpers-0.2.2-py3.6.egg/ocr4all_helpers/lib/nlbin.py", line 47, in adaptive_binarize
extreme = (np.sum(image<0.05)+np.sum(image>0.95))*1.0/np.prod(image.shape)
TypeError: '<' not supported between instances of 'Image' and 'float'

Is there any further information you need?

Regards,
Leonie

[maxnth]

Sorry to hear that the line segmentation isn't working as intended for you once again.
This looks like an easy to fix error in our helper scripts so I'm optimistic that we'll find a solution for this.
Would it be possible to send us one of the PAGE XML files where this error appears (either uploaded here or via mail)?

[lsubhd]

I hope the XML-file in the "data\processing" folder is what you wanted.

0017.zip

[maxnth]

I hope the XML-file in the "data\processing" folder is what you wanted.

My text probably was a bit misleading as I only mentioned the XML files but if possible, we'd need the binary / nrm image as well, as both annotations and images are used during line segmentation.

[lsubhd]

In that case - here is the image:

processing png -->

input png -->

[maxnth]

We sadly couldn't reproduce the error with the attached files but interestingly got an Image is not bi-level error instead, which is now fixed and will be shipped with the next minor release.
We'll keep looking into the above mentioned bug though.

[maxnth]

We changed parts of the line segmentation implementation and couldn't reproduce the error with that version as well.
In case the problem still persists, please reopen this issue.