forked from mawanda-jun/TableTrainNet
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathdataset_costants.py
37 lines (27 loc) · 900 Bytes
/
dataset_costants.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
import os
"""
All costants that are used to build the dataset are here.
PATH_TO_{} are those path in which the programs read
{}_TO_PATH are those in which the programs write
"""
DPI_EXTRACTION = 200
PATH_TO_IMAGES = os.path.dirname(os.path.abspath("dataset/Images"))
PATH_TO_ANNOTATIONS = os.path.dirname(os.path.abspath("dataset/Annotations"))
TRAIN_CSV_NAME = 'train_jpeg.csv'
TEST_CSV_NAME = 'test_jpeg.csv'
TRAIN_CSV_TO_PATH = os.path.abspath("data")
TEST_CSV_TO_PATH = os.path.abspath("data")
TF_TRAIN_RECORD_TO_PATH = os.path.abspath("data")
TF_TRAIN_RECORD_NAME = 'train_jpeg.record'
TF_TEST_RECORD_TO_PATH = os.path.dirname(os.path.abspath("data"))
TF_TEST_RECORD_NAME = 'test_jpeg.record'
ANNOTATIONS_EXTENSION = '.xml'
IMAGES_EXTENSION = '.jpeg'
TRAINING_PERCENTAGE = 0.995
TEST_PERCENTAGE = 0.005
MIN_WIDTH_BOX = 1
MIN_HEIGHT_BOX = 1
TABLE_DICT = {
'id': '1',
'name': 'table'
}