Merge branch 'binarization-threshold' into development

app/daemon/job_utils.py

@@ -65,6 +65,9 @@ def _create_job_service(job):
         command += f' -m {job.service_args["model"]}'
         if 'binarization' in job.service_args and job.service_args['binarization']:
             command += ' --binarize'
+        if 'ocropus_nlbin_threshold' in job.service_args and job.service_args['ocropus_nlbin_threshold']:
+            value = job.service_args['ocropus_nlbin_threshold']
+            command += f' --ocropus-nlbin-threshold {value}'
     elif job.service == 'transkribus-htr-pipeline':
         transkribus_htr_pipeline_model_id = job.service_args['model']
         command += f' -m {transkribus_htr_pipeline_model_id}'
@@ -144,8 +147,6 @@ def _create_job_service(job):
     )
     ''' ## Restart policy ## '''
     restart_policy = docker.types.RestartPolicy()
-    print(command)
-    print(mounts)
     try:
         docker_client.services.create(
             image,

app/services/forms.py

@@ -1,16 +1,12 @@
 from flask_login import current_user
 from flask_wtf import FlaskForm
 from flask_wtf.file import FileField, FileRequired
-from wtforms import (
-    BooleanField,
-    MultipleFileField,
-    SelectField,
-    StringField,
-    SubmitField,
-    ValidationError
-)
+from wtforms import (BooleanField, DecimalRangeField, MultipleFileField,
+                     SelectField, StringField, SubmitField, ValidationError)
 from wtforms.validators import InputRequired, Length
-from app.models import TesseractOCRPipelineModel, SpaCyNLPPipelineModel
+
+from app.models import SpaCyNLPPipelineModel, TesseractOCRPipelineModel
+
 from . import SERVICES
 
 
@@ -49,13 +45,16 @@ class CreateTesseractOCRPipelineJobForm(CreateJobBaseForm):
     binarization = BooleanField('Binarization')
     pdf = FileField('File', validators=[FileRequired()])
     model = SelectField('Model', validators=[InputRequired()])
+    ocropus_nlbin_threshold = DecimalRangeField(
+        render_kw={'min': 0, 'max': 1, 'step': 0.1, 'start': [0.5], 'disabled': True}
+    )
 
     def validate_binarization(self, field):
         service_info = SERVICES['tesseract-ocr-pipeline']['versions'][self.version.data]
         if field.data:
             if not('methods' in service_info and 'binarization' in service_info['methods']):
                 raise ValidationError('Binarization is not available')
-
+              
     def validate_pdf(self, field):
         if field.data.mimetype != 'application/pdf':
             raise ValidationError('PDF files only!')
@@ -68,10 +67,16 @@ class CreateTesseractOCRPipelineJobForm(CreateJobBaseForm):
         if self.binarization.render_kw is None:
             self.binarization.render_kw = {}
         self.binarization.render_kw['disabled'] = True
+        if self.ocropus_nlbin_threshold.render_kw is None:
+            self.ocropus_nlbin_threshold.render_kw = {}
+        self.ocropus_nlbin_threshold.render_kw['disabled'] = True
         if 'methods' in service_info:
             if 'binarization' in service_info['methods']:
                 if 'disabled' in self.binarization.render_kw:
                     del self.binarization.render_kw['disabled']
+            if 'ocropus_nlbin_threshold' in service_info['methods']:
+                if 'disabled' in self.ocropus_nlbin_threshold.render_kw:
+                    del self.ocropus_nlbin_threshold.render_kw['disabled']
         models = [
             x for x in TesseractOCRPipelineModel.query.order_by(TesseractOCRPipelineModel.title).all()
             if version in x.compatible_service_versions and (x.shared == True or x.user == current_user)

app/services/routes.py

@@ -79,7 +79,8 @@ def tesseract_ocr_pipeline():
                 service=service_name,
                 service_args={
                     'binarization': form.binarization.data,
-                    'model': hashids.decode(form.model.data)
+                    'model': hashids.decode(form.model.data),
+                    'ocropus_nlbin_threshold': float(form.ocropus_nlbin_threshold.data)
                 },
                 service_version=form.version.data,
                 user=current_user

app/services/services.yml

@@ -20,6 +20,7 @@ tesseract-ocr-pipeline:
     0.1.1:
       methods:
         - 'binarization'
+        - 'ocropus_nlbin_threshold'
       publishing_year: 2022
       url: 'https://gitlab.ub.uni-bielefeld.de/sfb1288inf/tesseract-ocr-pipeline/-/releases/v0.1.1'
 transkribus-htr-pipeline:

app/templates/services/tesseract_ocr_pipeline.html.j2

@@ -84,6 +84,11 @@
                   </label>
                 </div>
               </div>
+              <div class="col s12"><p>&nbsp;</p></div>  
+              <div class="col s9">
+                <p>Intensity (between 0 and 1)</p>
+                <p class="range-field">{{ form.ocropus_nlbin_threshold() }}</p>
+              </div>
               <div class="col s12"><p>&nbsp;</p></div>
               <div class="col s12 divider"></div>
               <div class="col s12"><p>&nbsp;</p></div>