-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathutil_tools.py
181 lines (146 loc) · 5.32 KB
/
util_tools.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
#-*- coding: utf-8 -*-
# STEP1
# 获取并显示USB摄像头实时视频
# 识别出人脸(OPENCV直接调用的API)
# STEP2
# 准备训练数据，对图像进行预处理与标注
# 用CNN读取数据集
import cv2
import sys
import numpy as np
import os
from PIL import Image
# GLOBAL VARIABLES
IMAGE_SIZE = 64
BLACK = [0,0,0] #RGB
IMAGES = [] #训练图像
LABELS = [] #训练图像的 label
PATH = 'F:\\Coursera\\FACEDETECTION_hzc\\datasets' #for small test
def CatchUsbVedio(window_name, camera_id,catch_pic_num,path_name):
cv2.namedWindow(window_name)
#视频来源
#1.保存的视频
#2.网络摄像头
cap = cv2.VideoCapture(camera_id)
# 告诉OpenCV使用人脸分类器
classifier = cv2.CascadeClassifier("F:/OPENCV/opencv/sources/data/haarcascades_cuda/haarcascade_frontalface_alt2.xml")
# 人脸的anchor box的颜色
color = (255,0,0)
num = 0
while cap.isOpened():
ok,frame = cap.read() #读取一帧
if not ok:
break
#将图像转为灰度图像
gray = cv2.cvtColor(frame,cv2.COLOR_BGR2GRAY)
#人脸检测
# ScaleFactor:缩放比例,同一物体，与相机距离不同，大小亦不相同，必须将其缩放到一定大小方便识别
# minNeighbors:对特征点周边有多少有效点同时检测，可避免选取的特征点太小导致易漏
# minSize:特征点的最小值
FaceRects = classifier.detectMultiScale(gray, scaleFactor=1.2, minNeighbors=3, minSize=(32,32))
if len(FaceRects)>0: #大于0检测出人脸
#输出多张人脸
for FaceRect in FaceRects: #单独框出每一个人的脸
x,y,w,h = FaceRect
#保存当前帧
img_name = '%s%d.jpg'%(path_name,num)
image = frame[y-10:y+h+10,x-10:x+w+10] #裁剪出人脸
cv2.imwrite(img_name, image)
num+=1
if num>(catch_pic_num):
break
#画出矩形框
cv2.rectangle(frame, (x-10,y-10), (x+w+10,y+h+10), (255,0,0), 2)
#显示当前捕捉了多少张人脸
font = cv2.FONT_HERSHEY_COMPLEX
cv2.putText(frame,'num:%d'%(num),(x+10,y+30),font,1,(255,0,255),4)
#超过最大保存数，结束程序
if num > (catch_pic_num):
break
#显示图像
#等待10ms,按'q'键退出
cv2.imshow(window_name, frame)
c = cv2.waitKey(10)
if c & 0xFF == ord('q'):
break
#释放摄像头并销毁所有窗口
cap.release()
cv2.destroyAllWindows()
# 调整图像尺寸大小 64x64加快学习
def resize_image(image,height=IMAGE_SIZE,width=IMAGE_SIZE):
top,bottom,left,right=(0,0,0,0)
#获取图像尺寸
h,w, _ = image.shape
#长短不一的，找到最长边
longest_edge = max(h,w)
#计算短边需要增加的像素，使其与长边相等
if h < longest_edge:
dh = longest_edge - h
top = dh//2
bottom = dh-top
elif w < longest_edge:
dw = longest_edge - w
left = dw//2
right = dw-left
else:
pass
# 给图像增加边界，使图像长宽等长，cv2.BORDER_CONSTANT指定边界颜色由value指定
# 给图像增加边界
# https://www.cnblogs.com/pakfahome/p/3914318.html
# BORDER_REFLICATE 直接用边界颜色填充
# BORDER_REFLECT 倒映
# BORDER_REFLECT_101 倒映 倒映时，会空开边界
# BORDER_WRAP 仿射
# BORDER_CONSTANT 填充颜色，颜色值由value定，这里用的是BLACK=[0,0,0]
# 比如200x300的>>64x64,需要填充黑边在上下
constant = cv2.copyMakeBorder(image,top,bottom,left,right,cv2.BORDER_CONSTANT,value=BLACK)
#调整图像大小并返回
return cv2.resize(constant,(height,width))
images=[]
labels=[]
def read_path(path_name):
for dir_item in os.listdir(path_name):
# 当前目录下所有文件
#从初始路径开始，合并成可识别的操作路径
# os.path.join将多个路径组合后返回，第一个绝对路径之前的参数将被忽略。
# full_path = os.path.abspath(os.path.join(path_name,dir_item))
full_path = os.path.join(path_name,dir_item)
# print(full_path) #当前路径
#如果是文件夹，继续递归调用
if os.path.isdir(full_path): #如果是文件夹，继续递归调用
read_path(full_path) #递归
else: #文件
if dir_item.endswith('.jpg'):
image = cv2.imread(full_path)
# print(image.shape)
# cv2.imshow("test",image)
# cv2.waitKey(10) #10ms
# cv2.destroyAllWindows()
image = resize_image(image,IMAGE_SIZE,IMAGE_SIZE) #调用函数进行处理
# 实际使用注释掉
# cv2.imwrite('1.jpg',image)
# cv2.imshow("test",image)
# cv2.waitKey(0)
# cv2.destroyAllWindows()
images.append(image)
labels.append(path_name)
return images,labels
#从指定路径读取训练数据
def load_dataset(path_name):
images,labels = read_path(path_name)
# print("Before ...",images.shape)
#将所有图片转为四维数组，尺寸为：(nums,image_size,image_size,channels)
images = np.array(images)
print('After ...',images.shape)
# print(labels)
#标注数据"HeZichen"是0, "Luna"是1
labels = np.array([0 if label.endswith("HeZichen") else 1 for label in labels])
return images,labels
# if __name__ == '__main__':
# 0 就是你的laptop的摄像头
# CatchUsbVedio("截取视频流", 0, 650, "datasets/HeZichen/")
# test = cv2.imread(PATH)
# cv2.imshow('image',test)
# cv2.waitKey(0)
# cv2.destroyAllWindows()
# images,labels = load_dataset(PATH)