-
Notifications
You must be signed in to change notification settings - Fork 0
/
Parsing.py
129 lines (113 loc) · 3.23 KB
/
Parsing.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
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
from functools import lru_cache
from json import load as load_json
from csv import reader as csv_reader
from typing import Literal, NamedTuple
class ImageInfo (NamedTuple):
id : int
width : int
height : int
fileName : str
# liscense_id : int
# coco_url : str
# date_captured : str
# flickr_url : str
class Category (NamedTuple):
id : int
name : str
superCategory : str
class Point (NamedTuple):
x : float
y : float
Polygons = list[list[Point]]
"Represents a list of polygons, each polygon is a list of Points"
class RLE (NamedTuple):
width : int
height : int
RLE :list[int]
class Rectangle (NamedTuple):
x : float
y : float
width : float
height : float
class ObjectSegmentation (NamedTuple):
id : int
image : ImageInfo
category : Category
boundingBox : Rectangle
isCrowd : bool
totalArea :float
segmentation : Polygons | RLE
@lru_cache(1)
def get_categories () -> dict[int,Category]:
FILE_PATH = "Data/ParsedAnnotations/CategoryList.csv"
retr = {}
with open(FILE_PATH) as fin:
fin.readline()
for line in csv_reader(fin, delimiter='\t'):
id = int(line[0])
retr[id] = Category(
id,
line[1],
line[2]
)
return retr
@lru_cache(1)
def get_image_list () -> dict[int,ImageInfo]:
FILE_PATH = "Data/ParsedAnnotations/ImageList.csv"
retr = {}
with open(FILE_PATH) as fin:
fin.readline()
for line in csv_reader(fin, delimiter='\t'):
id = int(line[0])
retr[id] = ImageInfo(
id,
int(line[1]),
int(line[2]),
line[3]
)
return retr
@lru_cache(2)
def get_segmentations (type:Literal["train","val"]) -> dict[int,Polygons|RLE]:
FILE_PATH = f"Data/ParsedAnnotations/Segmentations_{type}.json"
retr = {}
with open(FILE_PATH) as fin:
obj = load_json(fin)
for id, seg in obj.items():
id = int(id)
if isinstance(seg, dict):
size = seg["size"]
retr[id] = RLE(size[0], size[1], seg["counts"])
else:
retr[id] = Polygons(
[
Point(poly[i], poly[i+1])
for i in range(0, len(poly)-1, 2)
] for poly in seg
)
return retr
@lru_cache(2)
def get_objects (type:Literal["train","val"]) -> dict[int,ObjectSegmentation]:
cats = get_categories()
images = get_image_list()
segs = get_segmentations(type)
FILE_PATH = f"Data/ParsedAnnotations/Objects_{type}.csv"
retr = {}
with open(FILE_PATH) as fin:
fin.readline()
for line in csv_reader(fin, delimiter='\t'):
id = int(line[0])
retr[id] = ObjectSegmentation(
id,
images[int(line[1])], # image
cats[int(line[2])], # category
Rectangle(
float(line[4]),
float(line[5]),
float(line[6]),
float(line[7])
),
bool(line[8]), # isCrowd
float(line[3]),
segs[id]
)
return retr