diff options
author | Gustaf Rydholm <gustaf.rydholm@gmail.com> | 2022-09-27 01:44:49 +0200 |
---|---|---|
committer | Gustaf Rydholm <gustaf.rydholm@gmail.com> | 2022-09-27 01:44:49 +0200 |
commit | 9b8e14d89f0ef2508ed11f994f73af624155fe1d (patch) | |
tree | 10d9c138f4449646c8b5c0f95003ba86b831d04d /text_recognizer/metadata | |
parent | 63376b1c2da81c23ad5239f908b640cd42a514c7 (diff) |
Update data modules
Diffstat (limited to 'text_recognizer/metadata')
-rw-r--r-- | text_recognizer/metadata/iam_lines.py | 5 | ||||
-rw-r--r-- | text_recognizer/metadata/iam_paragraphs.py | 6 | ||||
-rw-r--r-- | text_recognizer/metadata/iam_synthetic_paragraphs.py | 4 |
3 files changed, 12 insertions, 3 deletions
diff --git a/text_recognizer/metadata/iam_lines.py b/text_recognizer/metadata/iam_lines.py index 73b2a28..a94041c 100644 --- a/text_recognizer/metadata/iam_lines.py +++ b/text_recognizer/metadata/iam_lines.py @@ -9,7 +9,12 @@ CHAR_WIDTH = emnist.INPUT_SHAPE[0] // IMAGE_SCALE_FACTOR # rough estimate IMAGE_HEIGHT = 112 // IMAGE_SCALE_FACTOR IMAGE_WIDTH = 3072 // IMAGE_SCALE_FACTOR # rounding up IAMLines empirical maximum width +SEED = 4711 DIMS = (1, IMAGE_HEIGHT, IMAGE_WIDTH) OUTPUT_DIMS = (89, 1) +IMAGE_HEIGHT = 56 +IMAGE_WIDTH = 1024 +MAX_LABEL_LENGTH = 89 +MAX_WORD_PIECE_LENGTH = 72 MAPPING = emnist.MAPPING diff --git a/text_recognizer/metadata/iam_paragraphs.py b/text_recognizer/metadata/iam_paragraphs.py index 70b802b..7bb909a 100644 --- a/text_recognizer/metadata/iam_paragraphs.py +++ b/text_recognizer/metadata/iam_paragraphs.py @@ -8,11 +8,13 @@ NEW_LINE_TOKEN = "\n" MAPPING = [*emnist.MAPPING, NEW_LINE_TOKEN] # must match IMAGE_SCALE_FACTOR for IAMLines to be compatible with synthetic paragraphs +SEED = 4711 IMAGE_SCALE_FACTOR = 2 +IMAGE_HEIGHT = 1152 // IMAGE_SCALE_FACTOR +IMAGE_WIDTH = 1280 // IMAGE_SCALE_FACTOR +MAX_LABEL_LENGTH = 682 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) diff --git a/text_recognizer/metadata/iam_synthetic_paragraphs.py b/text_recognizer/metadata/iam_synthetic_paragraphs.py index 15bcfc8..6c11d28 100644 --- a/text_recognizer/metadata/iam_synthetic_paragraphs.py +++ b/text_recognizer/metadata/iam_synthetic_paragraphs.py @@ -2,5 +2,7 @@ import text_recognizer.metadata.iam_paragraphs as iam_paragraphs import text_recognizer.metadata.shared as shared NEW_LINE_TOKEN = iam_paragraphs.NEW_LINE_TOKEN - +DATA_DIRNAME = shared.DATA_DIRNAME PROCESSED_DATA_DIRNAME = shared.DATA_DIRNAME / "processed" / "iam_synthetic_paragraphs" + +MAX_LABEL_LENGTH = 682 |