From 27ff7d113108e9cc51ddc5ff13b648b9c75fa865 Mon Sep 17 00:00:00 2001
From: Gustaf Rydholm <gustaf.rydholm@gmail.com>
Date: Tue, 27 Sep 2022 00:08:04 +0200
Subject: Add metadata

---
 text_recognizer/metadata/iam_paragraphs.py | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)
 create mode 100644 text_recognizer/metadata/iam_paragraphs.py

(limited to 'text_recognizer/metadata/iam_paragraphs.py')

diff --git a/text_recognizer/metadata/iam_paragraphs.py b/text_recognizer/metadata/iam_paragraphs.py
new file mode 100644
index 0000000..70b802b
--- /dev/null
+++ b/text_recognizer/metadata/iam_paragraphs.py
@@ -0,0 +1,18 @@
+import text_recognizer.metadata.emnist as emnist
+import text_recognizer.metadata.shared as shared
+
+
+PROCESSED_DATA_DIRNAME = shared.DATA_DIRNAME / "processed" / "iam_paragraphs"
+
+NEW_LINE_TOKEN = "\n"
+MAPPING = [*emnist.MAPPING, NEW_LINE_TOKEN]
+
+# must match IMAGE_SCALE_FACTOR for IAMLines to be compatible with synthetic paragraphs
+IMAGE_SCALE_FACTOR = 2
+IMAGE_HEIGHT, IMAGE_WIDTH = 576, 640
+IMAGE_SHAPE = (IMAGE_HEIGHT, IMAGE_WIDTH)
+
+MAX_LABEL_LENGTH = 682
+
+DIMS = (1, IMAGE_HEIGHT, IMAGE_WIDTH)
+OUTPUT_DIMS = (MAX_LABEL_LENGTH, 1)
-- 
cgit v1.2.3-70-g09d2