-
Notifications
You must be signed in to change notification settings - Fork 14
/
dataset.py
189 lines (150 loc) · 5.81 KB
/
dataset.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
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
"""
Author: Amr Elsersy
email: [email protected]
-----------------------------------------------------------------------------------
Description: FER2013 dataset
"""
import argparse
import cv2
from torch.utils.data import Dataset, DataLoader
import torchvision.transforms.transforms as transforms
import pandas as pd
import os
import numpy as np
import torch
from utils import get_label_emotion, normalization, histogram_equalization, standerlization, normalize_dataset_mode_1, normalize_dataset_mode_255, get_transforms
class FER2013(Dataset):
"""
FER2013 format:
index emotion pixels Usage
index: id of series
emotion: label (from 0 - 6)
pixels: 48x48 pixel value (uint8)
Usage: [Training, PrivateTest, PublicTest]
"""
def __init__(self, root='../data', mode = 'train', transform = None):
self.root = root
self.transform = transform
assert mode in ['train', 'val', 'test']
self.mode = mode
self.csv_path = os.path.join(self.root, 'fer2013.csv')
self.df = pd.read_csv(self.csv_path)
# print(self.df)
if self.mode == 'train':
self.df = self.df[self.df['Usage'] == 'Training']
elif self.mode == 'val':
self.df = self.df[self.df['Usage'] == 'PrivateTest']
else:
self.df = self.df[self.df['Usage'] == 'PublicTest']
def __getitem__(self, index: int):
data_series = self.df.iloc[index]
emotion = data_series['emotion']
pixels = data_series['pixels']
# to numpy
face = list(map(int, pixels.split(' ')))
face = np.array(face).reshape(48,48).astype(np.uint8)
if self.transform:
face = histogram_equalization(face)
# face = normalization(face)
face = self.transform(face)
return face, emotion
def __len__(self) -> int:
return self.df.index.size
def create_train_dataloader(root='../data', batch_size=64):
dataset = FER2013(root, mode='train', transform=get_transforms())
dataloader = DataLoader(dataset, batch_size, shuffle=True)
return dataloader
def create_val_dataloader(root='../data', batch_size=2):
dataset = FER2013(root, mode='val', transform=transforms.ToTensor())
dataloader = DataLoader(dataset, batch_size, shuffle=False)
return dataloader
def create_test_dataloader(root='../data', batch_size=1):
# transform = transforms.ToTensor()
transform = get_transforms()
dataset = FER2013(root, mode='test', transform=transform)
dataloader = DataLoader(dataset, batch_size, shuffle=False)
return dataloader
def calculate_dataset_mean_std(dataset:FER2013):
n = len(dataset)
means = []
stds = []
for i in range(n):
image, _ = dataset[i]
# image = image/ 255
mean = np.mean(image)
std = np.std(image)
means.append(mean)
stds.append(std)
print(f'i={i}, mean = {mean}, std = {std}')
mean = np.mean(means)
std = np.mean(stds)
print(f'\n\t Mean = {mean} ... Std = {std}\n')
def test_dataloader_main():
dataloader = create_test_dataloader()
for image, label in dataloader:
image = image.squeeze().numpy()
cv2.imshow('img', image)
print(image.shape)
if cv2.waitKey(0) == 27:
cv2.destroyAllWindows()
break
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('--mode', type=str, choices=['train', 'test', 'val'], default='train', help='dataset mode')
parser.add_argument('--datapath', type=str, default='../data')
parser.add_argument('--mean_std', action='store_true', help='calculate mean std of dataset')
parser.add_argument('--test', action='store_true', help='test augumentation')
args = parser.parse_args()
if args.test:
test_dataloader_main()
exit(0)
dataset = FER2013(args.datapath, args.mode)
print(f'dataset size = {len(dataset)}')
if args.mean_std:
calculate_dataset_mean_std(dataset)
exit(0)
for i in range(len(dataset)):
face, emotion = dataset[i]
# print('emotion',emotion)
# print('shape',face.shape)
face = np.copy(face)
print(f"before min:{np.min(face)}, max:{np.max(face)}, mean:{np.mean(face)}, std:{np.std(face)}")
face1 = normalize_dataset_mode_255(face)
# face1 = normalization(face)
face2 = standerlization(face)
print(f"after min:{np.min(face)}, max:{np.max(face)}, mean:{np.mean(face)}, std:{np.std(face)}\n")
face = cv2.resize(face, (200,200))
cv2.putText(face, get_label_emotion(emotion), (0,20), cv2.FONT_HERSHEY_COMPLEX, 1, (255,255,255))
cv2.imshow('face', face)
face1 = cv2.resize(face1, (200,200))
face2 = cv2.resize(face2, (200,200))
cv2.imshow('normalization', face1)
cv2.imshow('standerlization', face2)
if cv2.waitKey(0) == 27:
cv2.destroyAllWindows()
break
# df = pd.DataFrame({
# "name": ["amr",'ELSERSY', 'sersy'],
# 'salary': [100,20,3000],
# 'job': ['software', 'ray2', 'mech']
# }, index=[0,1,4])
# print(df.index.size)
# print(df)
# ser = df[df['salary'] > 50]
# print(type(ser)) # dataframe
# print(ser)
# salary = df[df['salary'] < 200] # dataframe
# print(type(salary.iloc[1])) # series
# print(salary.iloc[1],'\n')
# print(salary.iloc[0]['name'])
# print(salary.shape)
# print(salary.iloc[0]['name'])
# df = df[df['salary'] == 3000]
# print(df[['name', "salary"]])
# df = df[df['salary'] < 500]
# print(df.index)
# print(df.iloc[0].values)
# print(type(df.values))
# print('==========================')
# # print(df.count()) # count of each series in the dataframe
# print(df.iloc[0].count())