Merge pull request #52 from OCR-D/checkpoint_dir

Checkpoint dir

.gitignore

@@ -110,3 +110,4 @@ venv.bak/
 /actevedef_718448162*
 /repo
 /test/assets
+gt4histocr-calamari*

README.md

@@ -69,6 +69,13 @@ Finally recognize the text using ocrd_calamari and the downloaded model:
 ocrd-calamari-recognize -p '{ "checkpoint": "../gt4histocr-calamari1/*.ckpt.json" }' -I OCR-D-SEG-LINE -O OCR-D-OCR-CALAMARI
 ```
 
+or
+
+```
+ocrd-calamari-recognize -P checkpoint_dir ../gt4histocr-calamari1 -I OCR-D-SEG-LINE -O OCR-D-OCR-CALAMARI
+```
+
+
 You may want to have a look at the [ocrd-tool.json](ocrd_calamari/ocrd-tool.json) descriptions
 for additional parameters and default values.
 

ocrd_calamari/ocrd-tool.json

@@ -18,6 +18,10 @@
         "OCR-D-OCR-CALAMARI"
       ],
       "parameters": {
+        "checkpoint_dir": {
+          "description": "The directory containing calamari model files (*.ckpt.json). Uses all checkpoints in that directory",
+          "type": "string", "format": "file", "cacheable": true
+        },
         "checkpoint": {
           "description": "The calamari model files (*.ckpt.json)",
           "type": "string", "format": "file", "cacheable": true

ocrd_calamari/recognize.py

@@ -42,6 +42,8 @@ class CalamariRecognize(Processor):
 
     def _init_calamari(self):
 
+        if self.parameter.get('checkpoint_dir', None):
+            self.parameter['checkpoint'] = '%s/*.ckpt.json' % self.parameter['checkpoint_dir']
         checkpoints = glob(self.parameter['checkpoint'])
         self.predictor = MultiPredictor(checkpoints=checkpoints)
 

test/test_recognize.py

@@ -14,7 +14,8 @@ from .base import assets
 
 METS_KANT = assets.url_of('kant_aufklaerung_1784-page-region-line-word_glyph/data/mets.xml')
 WORKSPACE_DIR = '/tmp/test-ocrd-calamari'
-CHECKPOINT = os.path.join(os.getcwd(), 'gt4histocr-calamari1/*.ckpt.json')
+CHECKPOINT_DIR = os.path.join(os.getcwd(), 'gt4histocr-calamari1')
+CHECKPOINT = os.path.join(CHECKPOINT_DIR, '*.ckpt.json')
 
 # Because XML namespace versions are so much fun, we not only use one, we use TWO!
 NSMAP = { "pc": "http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15" }
@@ -81,6 +82,22 @@ def test_recognize(workspace):
     with open(page1, "r", encoding="utf-8") as f:
         assert "verſchuldeten" in f.read()
 
+def test_recognize_with_checkpoint_dir(workspace):
+    CalamariRecognize(
+        workspace,
+        input_file_grp="OCR-D-GT-SEG-LINE",
+        output_file_grp="OCR-D-OCR-CALAMARI",
+        parameter={
+            "checkpoint_dir": CHECKPOINT_DIR,
+        }
+    ).process()
+    workspace.save_mets()
+
+    page1 = os.path.join(workspace.directory, "OCR-D-OCR-CALAMARI/OCR-D-OCR-CALAMARI_0001.xml")
+    assert os.path.exists(page1)
+    with open(page1, "r", encoding="utf-8") as f:
+        assert "verſchuldeten" in f.read()
+
 
 def test_word_segmentation(workspace):
     CalamariRecognize(