-
Notifications
You must be signed in to change notification settings - Fork 6
/
Copy pathdataloader.py
136 lines (111 loc) · 4.71 KB
/
dataloader.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
130
131
132
133
134
135
136
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Wed Nov 2 18:13:15 2022
@author: kinsleyhsu
"""
import os
import numpy as np
import torch
from torch.utils.data import Dataset, DataLoader
from torchvision import transforms
from pandas import read_csv
from PIL import Image
from designSeq import reorder
def box_xyxy_to_cxcywh(x):
x0, y0, x1, y1 = x.unbind(-1)
b = [(x0 + x1) / 2, (y0 + y1) / 2,
(x1 - x0), (y1 - y0)]
return torch.stack(b, dim=-1)
class canvasLayout(Dataset):
def __init__(self, inp_dir, sal_dir_1, sal_dir_2, csv_path, max_elem=8):
img = os.listdir(inp_dir)
self.inp = list(map(lambda x: os.path.join(inp_dir, x), img))
self.sal_1 = list(map(lambda x: os.path.join(sal_dir_1, x.replace(".png", "_pred.png")), img))
self.sal_2 = list(map(lambda x: os.path.join(sal_dir_2, x), img))
df = read_csv(csv_path)
self.max_elem = max_elem
self.poster_name = list(map(lambda x: "train/" + x.replace("_mask", ""), img))
self.groups = df.groupby(df.poster_path)
self.transform = transforms.Compose([
transforms.Resize([350, 240]),
transforms.ToTensor()
])
def __len__(self):
return len(self.inp)
def __getitem__(self, idx):
img_inp = Image.open(self.inp[idx]).convert("RGB")
img_sal_1 = Image.open(self.sal_1[idx]).convert("L")
img_sal_2 = Image.open(self.sal_2[idx]).convert("L")
img_sal = Image.fromarray(np.maximum(np.array(img_sal_1), np.array(img_sal_2)))
img_inp = self.transform(img_inp)
img_sal = self.transform(img_sal)
cc = torch.concat([img_inp, img_sal])
label = np.zeros((self.max_elem, 2, 4))
sliced_df = self.groups.get_group(self.poster_name[idx])
cls = list(sliced_df["cls_elem"])
box = torch.tensor(list(map(eval, sliced_df["box_elem"])))
order = reorder(cls, box, "xyxy", self.max_elem)
for i in range(len(order)):
idx = order[i]
label[i][0][int(cls[idx])] = 1
label[i][1] = box[idx]
if label[i][1][0] > label[i][1][2] or label[i][1][1] > label[i][1][3]:
label[i][1][:2], label[i][1][2:] = label[i][1][2:], label[i][1][:2]
label[i][1] = box_xyxy_to_cxcywh(torch.tensor(label[i][1]))
label[i][1][::2] /= 513
label[i][1][1::2] /= 750
for i in range(len(order), self.max_elem):
label[i][0][0] = 1
return cc, torch.tensor(label).float()
class canvas(Dataset):
def __init__(self, bg_dir, sal_dir_1, sal_dir_2, train=True):
img = os.listdir(bg_dir)
if train:
img = img[:4]
else:
torch.save(img, "test_order.pt")
self.bg = list(map(lambda x: os.path.join(bg_dir, x), img))
self.sal_1 = list(map(lambda x: os.path.join(sal_dir_1, x.replace(".png", "_pred.png")), img))
self.sal_2 = list(map(lambda x: os.path.join(sal_dir_2, x), img))
self.transform = transforms.Compose([
transforms.Resize([350, 240]),
transforms.ToTensor()
])
def __len__(self):
return len(self.bg)
def __getitem__(self, idx):
img_bg = Image.open(self.bg[idx]).convert("RGB")
img_sal_1 = Image.open(self.sal_1[idx]).convert("L")
img_sal_2 = Image.open(self.sal_2[idx]).convert("L")
img_sal = Image.fromarray(np.maximum(np.array(img_sal_1), np.array(img_sal_2)))
img_bg = self.transform(img_bg)
img_sal = self.transform(img_sal)
cc = torch.concat([img_bg, img_sal])
return cc
def main():
test_batch_size = 4
test_bg_dir = "../Dataset/640_canvas"
test_sal_dir_1 = "../Dataset/640_sal"
test_sal_dir_2 = "../Dataset/640_sal_2"
train_batch_size = 32
train_inp_dir = "../Dataset/inp"
train_sal_dir_1 = "../Dataset/sal"
train_sal_dir_2 = "../Dataset/sal_2"
train_csv_path = "../Dataset/train_csv_9973.csv"
mex_elem = 8
testing_set = canvas(test_bg_dir, test_sal_dir_1, test_sal_dir_2)
testing_dl = DataLoader(testing_set, num_workers=16, batch_size=test_batch_size, shuffle=False)
img = next(iter(testing_dl))
print(img.shape)
training_set = canvasLayout(train_inp_dir, train_sal_dir_1, train_sal_dir_2, train_csv_path, mex_elem)
training_dl = DataLoader(training_set, num_workers=16, batch_size=train_batch_size, shuffle=True)
img, label = next(iter(training_dl))
print(img.shape, label.shape)
print(label[0])
if __name__ == "__main__":
import time
cur = time.time()
main()
end = time.time()
print(end-cur)