eScriptorium issueshttps://gitlab.inria.fr/scripta/escriptorium/-/issues2021-02-02T11:21:24+01:00https://gitlab.inria.fr/scripta/escriptorium/-/issues/239Old models2021-02-02T11:21:24+01:00daniel stoeklOld modelsNeither legacy segmentation models (I knew that) nor transcription models (this is new to me) are working in the new instance. Is there a workaround?Neither legacy segmentation models (I knew that) nor transcription models (this is new to me) are working in the new instance. Is there a workaround?https://gitlab.inria.fr/scripta/escriptorium/-/issues/237page display swallows some letters2020-05-26T09:29:56+02:00daniel stoeklpage display swallows some lettersIn this image (image and xml attached to next post) the system only displays some letters of a line. All letters are only displayed if you click on the line to transcribe it.
![image](/uploads/32f34ca9785247ad0aa012d3b79f60b7/image.png)
...In this image (image and xml attached to next post) the system only displays some letters of a line. All letters are only displayed if you click on the line to transcribe it.
![image](/uploads/32f34ca9785247ad0aa012d3b79f60b7/image.png)
It concerns both lines in the right column that begin later than the rest of the column (so that begin a little to the right to the rest).
![image](/uploads/bab639ac3960b62aa193639aef5db395/image.png)https://gitlab.inria.fr/scripta/escriptorium/-/issues/236downloading link needs to be changed2020-11-12T10:29:46+01:00daniel stoekldownloading link needs to be changedIf one downloads some pages and then downloads again some others, the green bar on top simply changes to "2 exports" but the link included in it is still the first link. It should be updated to the new link.If one downloads some pages and then downloads again some others, the green bar on top simply changes to "2 exports" but the link included in it is still the first link. It should be updated to the new link.https://gitlab.inria.fr/scripta/escriptorium/-/issues/234interactive polygonisation does not take other lines into consideration2021-02-02T11:35:17+01:00daniel stoeklinteractive polygonisation does not take other lines into considerationTo begin bug reports on the new instance :)
I segmented a page with the default model. Then had to join some oversegmented lines and the system miscalculated almost all polygons by simply disregarding potential borders to other lines
![...To begin bug reports on the new instance :)
I segmented a page with the default model. Then had to join some oversegmented lines and the system miscalculated almost all polygons by simply disregarding potential borders to other lines
![image](/uploads/c178bdbff1521e95c7b542a7f7063527/image.png).Uploading currently refuses attachment of page image. Will try to attach to next issue post.https://gitlab.inria.fr/scripta/escriptorium/-/issues/232empty coords in pagexml crashes import2020-05-13T18:34:22+02:00Robin Tissotempty coords in pagexml crashes importFound in cbad data![cPAS-2105](/uploads/81d36f533845753ee8c5dc747653b1bb/cPAS-2105.jpg)
```
<TextLine id="line_1546928226299_21956" custom="readingOrder {index:26;}">
<Coords points=""/>
<Ba...Found in cbad data![cPAS-2105](/uploads/81d36f533845753ee8c5dc747653b1bb/cPAS-2105.jpg)
```
<TextLine id="line_1546928226299_21956" custom="readingOrder {index:26;}">
<Coords points=""/>
<Baseline points="2115,2699 3485,2697"/>
</TextLine>
```
[cPAS-2105.xml](/uploads/fad2b7be4db8956fcc6beca0b3668921/cPAS-2105.xml)https://gitlab.inria.fr/scripta/escriptorium/-/issues/229transcription of rotated lines does not work as well as the others2020-12-17T15:43:29+01:00daniel stoekltranscription of rotated lines does not work as well as the othersThere is a bug somewhere in the pipeline that sends lines that are not in the usual horizontal rotation to kraken. Try transcribing the same page with the same model but rotate it 3 times by 90 degrees. The rotations will always give muc...There is a bug somewhere in the pipeline that sends lines that are not in the usual horizontal rotation to kraken. Try transcribing the same page with the same model but rotate it 3 times by 90 degrees. The rotations will always give much worse results. I sent this already by email but never got a response so I prefer to file it here so that it does not get lost.https://gitlab.inria.fr/scripta/escriptorium/-/issues/228Updating line order doesn't change next & previous of the line2021-03-04T10:52:29+01:00Robin TissotUpdating line order doesn't change next & previous of the lineSo clicking on the line to edit it and clicking the next or previous button leads to the wrong line (the one that was previously adjacent before reordering).So clicking on the line to edit it and clicking the next or previous button leads to the wrong line (the one that was previously adjacent before reordering).https://gitlab.inria.fr/scripta/escriptorium/-/issues/226pageXML import fail2020-09-24T11:49:37+02:00daniel stoeklpageXML import failI tried to import a couple of Sinai's pages. Some succeeded, others failed "cannot iterate over nonetype". Mayse bukh pages 6-10 worked, but pages 11-13 did not.
[Eyn_shoyn_mayśeh_bukh-11.xml](/uploads/c1ceb60630d2dfea6212e888c4ad0b71/E...I tried to import a couple of Sinai's pages. Some succeeded, others failed "cannot iterate over nonetype". Mayse bukh pages 6-10 worked, but pages 11-13 did not.
[Eyn_shoyn_mayśeh_bukh-11.xml](/uploads/c1ceb60630d2dfea6212e888c4ad0b71/Eyn_shoyn_mayśeh_bukh-11.xml)
![Eyn_shoyn_mayśeh_bukh-11](/uploads/71b362eae141fb13a9435cb6f7528206/Eyn_shoyn_mayśeh_bukh-11.png)https://gitlab.inria.fr/scripta/escriptorium/-/issues/225pageXML export change float to int2020-04-06T10:48:18+02:00daniel stoeklpageXML export change float to inteScriptorium produces float for Coord and Baseline, but cannot import that back. The import function is correct. The export function is not.
![image](/uploads/ab70cca854c97885bf4c9e9d15f1678c/image.png)eScriptorium produces float for Coord and Baseline, but cannot import that back. The import function is correct. The export function is not.
![image](/uploads/ab70cca854c97885bf4c9e9d15f1678c/image.png)https://gitlab.inria.fr/scripta/escriptorium/-/issues/224baselines in pagexml should be int not float2020-05-13T18:33:05+02:00daniel stoeklbaselines in pagexml should be int not floathttps://gitlab.inria.fr/scripta/escriptorium/-/issues/223order of zones in RTL documents2020-05-26T10:18:01+02:00daniel stoeklorder of zones in RTL documentsIt is very nice that one can correct the ordering of lines in multicolumn texts by associating them with a region. However, in a RTL text the columns should be counted from right to left, not as is now from left to right. The 'diplomatic...It is very nice that one can correct the ordering of lines in multicolumn texts by associating them with a region. However, in a RTL text the columns should be counted from right to left, not as is now from left to right. The 'diplomatic' panel should allow also the reorder them just as the lines, because I can think immediately of 3 million cases where this will be needed. keep up the great work!https://gitlab.inria.fr/scripta/escriptorium/-/issues/222weird error2020-05-26T09:31:00+02:00Robin Tissotweird errorFile "/usr/local/lib/python3.7/site-packages/django/core/handlers/exception.py" in inner
34. response = get_response(request)
File "/usr/local/lib/python3.7/site-packages/django/core/handlers/base.py" in _get_response
12...File "/usr/local/lib/python3.7/site-packages/django/core/handlers/exception.py" in inner
34. response = get_response(request)
File "/usr/local/lib/python3.7/site-packages/django/core/handlers/base.py" in _get_response
126. response = self.process_exception_by_middleware(e, request)
File "/usr/local/lib/python3.7/site-packages/django/core/handlers/base.py" in _get_response
124. response = wrapped_callback(request, *callback_args, **callback_kwargs)
File "/usr/local/lib/python3.7/site-packages/django/views/decorators/csrf.py" in wrapped_view
54. return view_func(*args, **kwargs)
File "/usr/local/lib/python3.7/site-packages/rest_framework/viewsets.py" in view
116. return self.dispatch(request, *args, **kwargs)
File "/usr/local/lib/python3.7/site-packages/rest_framework/views.py" in dispatch
495. response = self.handle_exception(exc)
File "/usr/local/lib/python3.7/site-packages/rest_framework/views.py" in handle_exception
455. self.raise_uncaught_exception(exc)
File "/usr/local/lib/python3.7/site-packages/rest_framework/views.py" in dispatch
492. response = handler(request, *args, **kwargs)
File "/usr/src/app/apps/api/views.py" in bulk_update
177. serializer.save()
File "/usr/local/lib/python3.7/site-packages/rest_framework/serializers.py" in save
726. self.instance = self.update(self.instance, validated_data)
File "/usr/src/app/apps/api/serializers.py" in update
124. ret.append(self.child.update(line, data))
File "/usr/local/lib/python3.7/site-packages/rest_framework/serializers.py" in update
974. info = model_meta.get_field_info(instance)
File "/usr/local/lib/python3.7/site-packages/rest_framework/utils/model_meta.py" in get_field_info
35. opts = model._meta.concrete_model._meta
Exception Type: AttributeError at /api/documents/322/parts/21904/lines/bulk_update/
Exception Value: 'NoneType' object has no attribute '_meta'
Request information:
USER: elena.lollihttps://gitlab.inria.fr/scripta/escriptorium/-/issues/221image deletion2020-04-06T10:49:05+02:00daniel stoeklimage deletionI uploaded an image of a text. Then I rotated the same image 180 degrees on my computer and uploaded it into the same document under the same name. This morning, both images were turned around 180 degrees. I then deleted both images from...I uploaded an image of a text. Then I rotated the same image 180 degrees on my computer and uploaded it into the same document under the same name. This morning, both images were turned around 180 degrees. I then deleted both images from the document and uploaded the correct first image again. It was automatically rotated by 180 degrees. document 349. Image attached. ![BL_seder](/uploads/0a8fd611bb23837fbaf013197498cb52/BL_seder.PNG)https://gitlab.inria.fr/scripta/escriptorium/-/issues/218destructive zoom2020-03-27T12:35:12+01:00daniel stoekldestructive zoomif you zoom in or out and then try to go to another image with the arrows or pgdown, the manuscript image does not get updated in the segmentation panel or the manuscript panel. The transcription panel gets updated.if you zoom in or out and then try to go to another image with the arrows or pgdown, the manuscript image does not get updated in the segmentation panel or the manuscript panel. The transcription panel gets updated.https://gitlab.inria.fr/scripta/escriptorium/-/issues/217Segtrain2020-05-05T08:37:01+02:00daniel stoeklSegtrainWhat exactly is happening in segtrain? I tried to train a model from Scratch on roughly 300 images. The model page gives 66% as a result after 11 epochs and stops there. And the results are terrible.
![image](/uploads/ef5c95b0732e8f2bbec...What exactly is happening in segtrain? I tried to train a model from Scratch on roughly 300 images. The model page gives 66% as a result after 11 epochs and stops there. And the results are terrible.
![image](/uploads/ef5c95b0732e8f2bbeca0e866d6400f5/image.png)
I train 11 epochs with the same material on the gpu and the results are ok, not fantastic but workable:
![image](/uploads/670498493a8c1a5736e348548f4e4851/image.png)
1. We know that the automatic evaluation does not correspond to manual evaluation. Could we rather do a fix number of epochs and save every 10th or every 50th model as Ben and I are doing on our gpus?
2. Put as little as possible into the testset in order to keep the trainingset as big as possible.https://gitlab.inria.fr/scripta/escriptorium/-/issues/216modelname if train on top of old model2020-04-02T14:34:38+02:00daniel stoeklmodelname if train on top of old modelIf one trains on top of an already trained model, the new model is not taken into account, but the new model bears the same name as the one on which one has trained. I also have the suspicion that when one applies the new model to some o...If one trains on top of an already trained model, the new model is not taken into account, but the new model bears the same name as the one on which one has trained. I also have the suspicion that when one applies the new model to some other pages the system does not take into account the new model but takes the old model instead.https://gitlab.inria.fr/scripta/escriptorium/-/issues/213eScriptorium version number missing2020-03-19T14:33:37+01:00daniel stoekleScriptorium version number missingCurrent opening screen shows
Version passthistobuildcmd
kraken, version 3.0.0.0b4.dev12Current opening screen shows
Version passthistobuildcmd
kraken, version 3.0.0.0b4.dev12https://gitlab.inria.fr/scripta/escriptorium/-/issues/212regions cannot be deleted2020-03-18T14:50:16+01:00daniel stoeklregions cannot be deletedI have a case of unsuppressable lines here: document/340/part/26883/edit/
You can play with this image but please do not change any other annotation.I have a case of unsuppressable lines here: document/340/part/26883/edit/
You can play with this image but please do not change any other annotation.https://gitlab.inria.fr/scripta/escriptorium/-/issues/210joining vertical lines bug2020-03-16T13:38:26+01:00daniel stoekljoining vertical lines bugJoining vertical lines is buggy.
It works with some combinations of line directions and angles but not with others.
I think it works for
a) top down lines where the top line has a direction between 12 and 3 o'clock.
It fails for
a) to...Joining vertical lines is buggy.
It works with some combinations of line directions and angles but not with others.
I think it works for
a) top down lines where the top line has a direction between 12 and 3 o'clock.
It fails for
a) top down lines where the top line has a direction between 9 to 12 o'clock
b) bottom up lines where the top line has a direction between 12 and o'clock.
c) bottom up lines where the top line has a direction between 9 to 12 o'clockhttps://gitlab.inria.fr/scripta/escriptorium/-/issues/209Exports do not include empty regions2020-11-10T16:17:51+01:00daniel stoeklExports do not include empty regionsI created a second region but it exported only 1.I created a second region but it exported only 1.