eScriptorium issueshttps://gitlab.inria.fr/scripta/escriptorium/-/issues2021-09-01T08:27:06+02:00https://gitlab.inria.fr/scripta/escriptorium/-/issues/356freeze automatic reordering button needed urgently2021-09-01T08:27:06+02:00daniel stoeklfreeze automatic reordering button needed urgentlyVery frequently we encounter the situation that complex line ordering has been manually fixed and then someone notices that a line needs to be added or deleted and all the reordering work needs to be redone. The best solution would be a ...Very frequently we encounter the situation that complex line ordering has been manually fixed and then someone notices that a line needs to be added or deleted and all the reordering work needs to be redone. The best solution would be a button on top of the segmentation panel (because here is what causes the reordering) that can toggle the ability to automatically reorder. This should be written into the dbase for the relative part and changed only if the button is hit another time to untoggle. It is really important to keep the workflow straight. Sometimes time for manual reordering correction can be longer than for all other steps.https://gitlab.inria.fr/scripta/escriptorium/-/issues/355Transcription lost (?), unable to download even the corrected segmentation2021-06-16T09:42:46+02:00Thibault ClériceTranscription lost (?), unable to download even the corrected segmentationI don't know if this is a bug or a feature, but here is what happened to two colleagues over the last couple days: they tried a lot of different segmenter, and ended up using it on manuscripts that had some ground truth of segmentation.
...I don't know if this is a bug or a feature, but here is what happened to two colleagues over the last couple days: they tried a lot of different segmenter, and ended up using it on manuscripts that had some ground truth of segmentation.
It seemed it removed their ability to transcribe (`e.line.currentTrans` is undefined` in console), and because there was no transcription, the export button would not appear (even so there is segmentation ground truth).
I found a "quick fix hack" that is: simply run a transcription model on ONE of the page at least. Button reappears, everybody is happy.
But I thought it might be a bug ?https://gitlab.inria.fr/scripta/escriptorium/-/issues/354transcription of same image changes when rotated2021-05-19T19:17:14+02:00daniel stoekltranscription of same image changes when rotatedE.g. normal: ![image](/uploads/7137fd166665c1e7eb7bb4558e932a48/image.png)
upside down: (screenshot exported to paint, rotated there and switched panel 2 and 3 to make it comparable): ![image](/uploads/65570cc744a0df4358d6a6336c0a51a6/im...E.g. normal: ![image](/uploads/7137fd166665c1e7eb7bb4558e932a48/image.png)
upside down: (screenshot exported to paint, rotated there and switched panel 2 and 3 to make it comparable): ![image](/uploads/65570cc744a0df4358d6a6336c0a51a6/image.png)
look especially on the short lines on the right or the completely empty lines.https://gitlab.inria.fr/scripta/escriptorium/-/issues/353Use a Many-to-Many relation between Document and OCRModel2021-05-21T16:16:40+02:00Bastien AbadieUse a Many-to-Many relation between Document and OCRModel
The goal is to convert the `OcrModel.document` Foreign Key to an `OcrModel.documents` Many-to-Many relation towards `Document` instances.
This means several low-level changes:
- an `OcrModel` would then be usable across multiple docume...
The goal is to convert the `OcrModel.document` Foreign Key to an `OcrModel.documents` Many-to-Many relation towards `Document` instances.
This means several low-level changes:
- an `OcrModel` would then be usable across multiple documents
- a document would still be usable across multiple `OcrModel` instance
- the `OcrModel.owner` user would be the administrator of the model, no access rights could be easily extracted/aggregated from documents
A M2M "through" model named `OcrModelDocument` would be created, only to store the relationship between `OcrModel` and `Document` alongside a `created` date (it's easier to do that from the beginning than a later migration to add it).
The `OcrModelDocument` model would also store boolean states descritbing the interactions between an `OcrModel` and a `Document`:
- `trained_on` would be used when the model has been trained on the document
- `executed_on` would be used when the model has been executed on the document to produce results
A unique constraing on `document_id` and `model_id` would ensure than only one M2M relation is possible between a model and a document.
This kind of changes can break a lot of functionalities in small ways, so we would need to fix the behaviour of (at least) these parts:
- document creation interface
- training tasks
- access control to models and documents
- API `OcrModelViewSet` and its usage
At this stage, we only update the database and surrounding code, reproducing current behaviour.
The resulting code should be mergeable as-is and serve as basis for features development.Eva BardouEva Bardouhttps://gitlab.inria.fr/scripta/escriptorium/-/issues/352increase/decrease fontsize button in transcription modal2021-05-17T00:25:46+02:00daniel stoeklincrease/decrease fontsize button in transcription modalCould we pleeeeeeeeeeeeeeeeeeeeeeeeeeeease get two buttons for increase/decrease font size for the comparison visualization in the transcription modal? It can be so tiny. I know the trick to switch the rem in the console but it is a nigh...Could we pleeeeeeeeeeeeeeeeeeeeeeeeeeeease get two buttons for increase/decrease font size for the comparison visualization in the transcription modal? It can be so tiny. I know the trick to switch the rem in the console but it is a nightmare to teach this in a tutorial and quite time consuming because one has to do it again as soon as one goes to the image wheel and back. Here is a common example display:
![image](/uploads/9a18924dcbd932b68dc97f388fe546bb/image.png)https://gitlab.inria.fr/scripta/escriptorium/-/issues/351newly created line in segmentation filled with wrong text2021-05-15T11:07:00+02:00daniel stoeklnewly created line in segmentation filled with wrong textIf I create a new line in the segmentation panel in a text with existing transcription and then want to add it's text, it is filled with the text of the previous/next line and I am not sure what would happen if I started to correct this....If I create a new line in the segmentation panel in a text with existing transcription and then want to add it's text, it is filled with the text of the previous/next line and I am not sure what would happen if I started to correct this. After hitting F5 it is fine, but this seems a digression.https://gitlab.inria.fr/scripta/escriptorium/-/issues/350Move api's bulk_create, bulk_update and bulk_delete to a single bulk endpoint2021-05-04T13:49:35+02:00Robin TissotMove api's bulk_create, bulk_update and bulk_delete to a single bulk endpointFor both lines and transcriptions. Use the request method to instantiate the serializer accordingly.For both lines and transcriptions. Use the request method to instantiate the serializer accordingly.https://gitlab.inria.fr/scripta/escriptorium/-/issues/349adding text to lines without polygons in either 3rd or 4th panel gives js error2021-05-04T11:29:45+02:00Robin Tissotadding text to lines without polygons in either 3rd or 4th panel gives js errorhttps://gitlab.inria.fr/scripta/escriptorium/-/issues/348Blank edit page when the title of the document contains '2021-05-04T12:32:00+02:00Robin TissotBlank edit page when the title of the document contains 'https://gitlab.inria.fr/scripta/escriptorium/-/issues/347Applying segmentation model fails2021-04-26T15:27:59+02:00Hayim LapinApplying segmentation model failshttps://www.escriptorium.fr/document/581/edit/
elements 6-10, using model: paris_BNF328_329trial_3__best several pages failed. Any way of telling whether this is a faulty model or a bug on the server side?
Document has the same 5 jpegs r...https://www.escriptorium.fr/document/581/edit/
elements 6-10, using model: paris_BNF328_329trial_3__best several pages failed. Any way of telling whether this is a faulty model or a bug on the server side?
Document has the same 5 jpegs repeated for comparison of segmentationhttps://gitlab.inria.fr/scripta/escriptorium/-/issues/346Server not updating from web app.2021-04-28T11:40:52+02:00Hayim LapinServer not updating from web app.Firefox on Windows.
* Not recalculating masks unless individual line is double clicked
* Not saving manual resegmentation
Could this just be a load issue on the server end?
Edit: also not renumbering lines after associating with regions.Firefox on Windows.
* Not recalculating masks unless individual line is double clicked
* Not saving manual resegmentation
Could this just be a load issue on the server end?
Edit: also not renumbering lines after associating with regions.https://gitlab.inria.fr/scripta/escriptorium/-/issues/345Add default model in list of models available for finetuning2021-04-26T12:23:50+02:00Alix ChaguéAdd default model in list of models available for finetuningAt the moment it is not possible to finetune the default segmentation model (blla.mlmodel) when we train a segmenter. Is it possible to add it to the list of available models?
![image](/uploads/2adcea4fe6b972c8843217abab2cdef6/image.png)At the moment it is not possible to finetune the default segmentation model (blla.mlmodel) when we train a segmenter. Is it possible to add it to the list of available models?
![image](/uploads/2adcea4fe6b972c8843217abab2cdef6/image.png)https://gitlab.inria.fr/scripta/escriptorium/-/issues/344Hitting ctrl+arrow in the transcription modal doesn't save2021-04-26T16:37:22+02:00Robin TissotHitting ctrl+arrow in the transcription modal doesn't savehttps://gitlab.inria.fr/scripta/escriptorium/-/issues/343too sensitive2021-04-06T01:54:09+02:00daniel stoekltoo sensitiveThere is some change in the behavior for mouse clicks in the segmentation panel. I have much pain in correcting layout because it starts creating lines all the time when I try to select a line instead. and then the escape key usually doe...There is some change in the behavior for mouse clicks in the segmentation panel. I have much pain in correcting layout because it starts creating lines all the time when I try to select a line instead. and then the escape key usually doesnt kill the unborn line right away.https://gitlab.inria.fr/scripta/escriptorium/-/issues/342Letters cut off in some scripts2021-04-19T16:42:41+02:00Robin TissotLetters cut off in some scriptsThe letters are not completely visible in the transcription modal input:
![image](/uploads/88faf3dd57a717d7a5331ac78d683de9/image.png)
![Zavia_270](/uploads/9c5f057f28bcc3010fecaa53c21c6065/Zavia_270.jpg)[Zavia_270.xml](/uploads/b4c84f...The letters are not completely visible in the transcription modal input:
![image](/uploads/88faf3dd57a717d7a5331ac78d683de9/image.png)
![Zavia_270](/uploads/9c5f057f28bcc3010fecaa53c21c6065/Zavia_270.jpg)[Zavia_270.xml](/uploads/b4c84f4d895834e4cd372a9e7dce8fd2/Zavia_270.xml)https://gitlab.inria.fr/scripta/escriptorium/-/issues/341Improving names of exported archives2021-03-23T18:09:18+01:00Alix ChaguéImproving names of exported archivesBecause there is no limit to the length of documents' names, and because some of our users named their documents with a lot of details, we are sometimes facing cases where the exported file creates a Windows bug where the name of the fil...Because there is no limit to the length of documents' names, and because some of our users named their documents with a lot of details, we are sometimes facing cases where the exported file creates a Windows bug where the name of the file is too long to be able to be manipulated once it's downloaded. Besides, it is very helpful to rely of document's ids, since we can easily reconstruct the url based on this id, which is not the case with the document's name (not unique nor permanent). So, I see 2 ways to impove the naming of the archive files exported from escriptorium:
- in [`export_%s_%s_%s.txt" % (slugify(document.name).replace('-', '_'),file_format, datetime.now().strftime('%Y%m%d%H%M'))`]( https://gitlab.inria.fr/scripta/escriptorium/-/blob/develop/app/apps/imports/tasks.py#L85) ; set a max length to `document.name`
- add the document's id in the name such as : `export_%s_%s_%s_%s.txt" % (document.pk, slugify(document.name).replace('-', '_'),file_format, datetime.now().strftime('%Y%m%d%H%M'))`
---
As a side comment, Transkribus enables the user to set the pattern of the exported filename. I have never used it so I consider it's probably too much effort to build an equivalent option. But I mention it still, juste in case. It looks like this:
![image](/uploads/dfbd462cf81a08894cdf978fad2bbf16/image.png)https://gitlab.inria.fr/scripta/escriptorium/-/issues/340Error while editing parts of documents when the document name contains a sing...2021-03-23T17:50:32+01:00TADJO TAKIANPI Yves ConstantinError while editing parts of documents when the document name contains a single quote.Input :
- Document name : DAHN d'Estournelles
- Some images
clicking on the edit tab displays nothing.
![Console_javascript_error](/uploads/3c37f3a10589e712ee16af38aa046139/Console_javascript_error.png)Input :
- Document name : DAHN d'Estournelles
- Some images
clicking on the edit tab displays nothing.
![Console_javascript_error](/uploads/3c37f3a10589e712ee16af38aa046139/Console_javascript_error.png)https://gitlab.inria.fr/scripta/escriptorium/-/issues/339copying empty lines in panel 4 should insert CR2021-05-04T12:31:51+02:00daniel stoeklcopying empty lines in panel 4 should insert CRCurrently if one copies from one transcription level into another and there are empty lines in the middle none of these empty lines is represented by a CR and all subsequent lines are copied into wrong lines. Also for text export, I woul...Currently if one copies from one transcription level into another and there are empty lines in the middle none of these empty lines is represented by a CR and all subsequent lines are copied into wrong lines. Also for text export, I would greatly support inserting empty lines with a CR rather than nothing.https://gitlab.inria.fr/scripta/escriptorium/-/issues/338Deleting transcriptions problems2021-05-25T17:49:11+02:00Robin TissotDeleting transcriptions problemsIt was impossible to delete the 'manual' transcription before as part of the code assume there should be at least one transcription, it seems that in some case it is possible now.
The deletion confirmation doesn't make it clear that it...It was impossible to delete the 'manual' transcription before as part of the code assume there should be at least one transcription, it seems that in some case it is possible now.
The deletion confirmation doesn't make it clear that it deletes for the whole document, not just this page.https://gitlab.inria.fr/scripta/escriptorium/-/issues/337keep line numbers small2021-03-05T22:24:12+01:00daniel stoeklkeep line numbers smallsometimes the line ordering algo gives wrong line numbers and it is necessary to zoom in to understand what is before what but the line numbers get so big one cannot see nothing. Would it be possible to keep them as small as they were in...sometimes the line ordering algo gives wrong line numbers and it is necessary to zoom in to understand what is before what but the line numbers get so big one cannot see nothing. Would it be possible to keep them as small as they were initially?
![image](/uploads/0d54754d5b5a4e08dc7798cc6ac04f6b/image.png)