blob: 70b802bcfbdee3cade379720150ea3145409cd4e (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
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)
|