drop obsolete multi-model binarization

src/eynollah/model_zoo/default_specs.py

@@ -45,42 +45,6 @@ DEFAULT_MODEL_SPECS = EynollahModelSpecSet([
         type='Keras',
     ),
 
-    EynollahModelSpec(
-        category="binarization_multi_1",
-        variant='',
-        filename="models_eynollah/eynollah-binarization-multi_2020_01_16/model_bin1",
-        dist_url=dist_url("binarization"),
-        dists=['binarization'],
-        type='Keras',
-    ),
-
-    EynollahModelSpec(
-        category="binarization_multi_2",
-        variant='',
-        filename="models_eynollah/eynollah-binarization-multi_2020_01_16/model_bin2",
-        dist_url=dist_url("binarization"),
-        dists=['binarization'],
-        type='Keras',
-    ),
-
-    EynollahModelSpec(
-        category="binarization_multi_3",
-        variant='',
-        filename="models_eynollah/eynollah-binarization-multi_2020_01_16/model_bin3",
-        dist_url=dist_url("binarization"),
-        dists=['binarization'],
-        type='Keras',
-    ),
-
-    EynollahModelSpec(
-        category="binarization_multi_4",
-        variant='',
-        filename="models_eynollah/eynollah-binarization-multi_2020_01_16/model_bin4",
-        dist_url=dist_url("binarization"),
-        dists=['binarization'],
-        type='Keras',
-    ),
-
     EynollahModelSpec(
         category="col_classifier",
         variant='',

src/eynollah/ocrd_cli_binarization.py

@@ -40,7 +40,7 @@ class SbbBinarizeProcessor(Processor):
         # resolve relative path via OCR-D ResourceManager
         assert isinstance(self.parameter, frozendict)
         model_zoo = EynollahModelZoo(basedir=self.parameter['model'])
-        self.binarizer = SbbBinarizer(model_zoo=model_zoo, mode='single', logger=self.logger)
+        self.binarizer = SbbBinarizer(model_zoo=model_zoo, logger=self.logger)
 
     def process_page_pcgts(self, *input_pcgts: Optional[OcrdPage], page_id: Optional[str] = None) -> OcrdPageResult:
         """
@@ -103,7 +103,7 @@ class SbbBinarizeProcessor(Processor):
                 line_image_bin = cv2pil(self.binarizer.run(image=pil2cv(line_image), use_patches=True))
                 # update PAGE (reference the image file):
                 line_image_ref = AlternativeImageType(comments=line_xywh['features'] + ',binarized')
-                line.add_AlternativeImage(region_image_ref)
+                line.add_AlternativeImage(line_image_ref)
                 result.images.append(OcrdPageResultImage(line_image_bin, line.id + '.IMG-BIN', line_image_ref))
 
         return result

src/eynollah/sbb_binarize.py

@@ -33,12 +33,10 @@ class SbbBinarizer:
         self,
         *,
         model_zoo: EynollahModelZoo,
-        mode: str,
         logger: Optional[logging.Logger] = None,
     ):
         self.logger = logger if logger else logging.getLogger('eynollah.binarization')
-        self.model_zoo = model_zoo
+        self.models = (model_zoo.model_path('binarization'), model_zoo.load_model('binarization'))
-        self.models = self.setup_models(mode)
         self.session = self.start_new_session()
 
     def start_new_session(self):
@@ -49,12 +47,6 @@ class SbbBinarizer:
         tensorflow_backend.set_session(session)
         return session
     
-    def setup_models(self, mode: str) -> Dict[Path, AnyModel]:
-        return {
-            self.model_zoo.model_path(v): self.model_zoo.load_model(v)
-            for v in (['binarization'] if mode == 'single' else [f'binarization_multi_{i}' for i in range(1, 5)])
-        }
-
     def end_session(self):
         tensorflow_backend.clear_session()
         self.session.close()
@@ -330,21 +322,21 @@ class SbbBinarizer:
             if image_path is not None:
                 image = cv2.imread(image_path)
             img_last = 0
-            for n, (model_file, model) in enumerate(self.models.items()):
+            model_file, model = self.models
-                self.logger.info('Predicting %s with model %s [%s/%s]', image_path if image_path else '[image]', model_file, n + 1, len(self.models.keys()))
+            self.logger.info('Predicting %s with model %s [%s/%s]', image_path if image_path else '[image]', model_file)
-                res = self.predict(model, image, use_patches)
+            res = self.predict(model, image, use_patches)
 
-                img_fin = np.zeros((res.shape[0], res.shape[1], 3))
+            img_fin = np.zeros((res.shape[0], res.shape[1], 3))
-                res[:, :][res[:, :] == 0] = 2
+            res[:, :][res[:, :] == 0] = 2
-                res = res - 1
+            res = res - 1
-                res = res * 255
+            res = res * 255
-                img_fin[:, :, 0] = res
+            img_fin[:, :, 0] = res
-                img_fin[:, :, 1] = res
+            img_fin[:, :, 1] = res
-                img_fin[:, :, 2] = res
+            img_fin[:, :, 2] = res
 
-                img_fin = img_fin.astype(np.uint8)
+            img_fin = img_fin.astype(np.uint8)
-                img_fin = (res[:, :] == 0) * 255
+            img_fin = (res[:, :] == 0) * 255
-                img_last = img_last + img_fin
+            img_last = img_last + img_fin
 
             kernel = np.ones((5, 5), np.uint8)
             img_last[:, :][img_last[:, :] > 0] = 255
@@ -361,22 +353,21 @@ class SbbBinarizer:
                 self.logger.info('Binarizing [%3d/%d] %s', i + 1, len(ls_imgs), image_name)
                 image = cv2.imread(os.path.join(dir_in,image_name) )
                 img_last = 0
-                for n, (model_file, model) in enumerate(self.models.items()):
+                model_file, model = self.models
-                    self.logger.info('Predicting %s with model %s [%s/%s]', image_name, model_file, n + 1, len(self.models.keys()))
+                self.logger.info('Predicting %s with model %s [%s/%s]', image_path if image_path else '[image]', model_file)
+                res = self.predict(model, image, use_patches)
 
-                    res = self.predict(model, image, use_patches)
+                img_fin = np.zeros((res.shape[0], res.shape[1], 3))
+                res[:, :][res[:, :] == 0] = 2
+                res = res - 1
+                res = res * 255
+                img_fin[:, :, 0] = res
+                img_fin[:, :, 1] = res
+                img_fin[:, :, 2] = res
 
-                    img_fin = np.zeros((res.shape[0], res.shape[1], 3))
+                img_fin = img_fin.astype(np.uint8)
-                    res[:, :][res[:, :] == 0] = 2
+                img_fin = (res[:, :] == 0) * 255
-                    res = res - 1
+                img_last = img_last + img_fin
-                    res = res * 255
-                    img_fin[:, :, 0] = res
-                    img_fin[:, :, 1] = res
-                    img_fin[:, :, 2] = res
-
-                    img_fin = img_fin.astype(np.uint8)
-                    img_fin = (res[:, :] == 0) * 255
-                    img_last = img_last + img_fin
 
                 kernel = np.ones((5, 5), np.uint8)
                 img_last[:, :][img_last[:, :] > 0] = 255