opencv计算机视觉学习笔记八
第九章 基于opencv的神经网络简介
1 人工神精网络ann
2 人工神精网络的结构
输入层
网络的输入数目
如动物有体重,长度,牙齿三个属性,网络则需要三个输入节点
中间层
输出层
与定义的类别数相同,如定义了猪,狗,猫,鸡,则输出层的数目为4
创建ANN常见规则
神经元数 位于输入/输出层之间, 接近输出层
较小的输入,神经元数=(输入+输出)/3*2
学习算法:
监督学习
非监督学习
强化学习
3 opencv中的ann
示例代码如下:
import cv2 import numpy as np # 创建ann,MLP 是multilayer perceptron 感知器 ann = cv2.ml.ANN_MLP_create() # 设置拓扑结构,通过数组来定义各层大小,分别对应输入/隐藏/输出 ann.setLayerSizes(np.array([9, 5, 9], dtype=np.uint8)) # 采用反向传播方式,还有一种方式ANN_MLP_RPROP,只有在有监督学习中才可以设置 ann.setTrainMethod(cv2.ml.ANN_MLP_BACKPROP) # 有点类似于向量机svm的 train函数 ann.train(np.array([[1.2, 1.3, 1.9, 2.2, 2.3, 2.9, 3.0, 3.2, 3.3]], dtype=np.float32), # 对应9个输入数据 cv2.ml.ROW_SAMPLE, # 如果提供以下几个参数就是有监督学习 np.array([[0, 0, 0, 0, 0, 1, 0, 0, 0]], dtype=np.float32)) # 输出层大小为9 print(ann.predict(np.array([[1.4, 1.5, 1.2, 2., 2.5, 2.8, 3., 3.1, 3.8]], dtype=np.float32))) # 输出结果为: # (5.0, #类标签 # array([[-0.06419383, -0.13360272, -0.1681568 , -0.18708915, 0.0970564 , #输入数据属于每个类的概率 # 0.89237726, 0.05093023, 0.17537238, 0.13388439]], dtype=float32))
基于ann的动物分类
示例代码如下:
import cv2 import numpy as np from random import randint # 创建ann animals_net = cv2.ml.ANN_MLP_create() # 设定train函数为弹性反向传播 animals_net.setTrainMethod(cv2.ml.ANN_MLP_RPROP | cv2.ml.ANN_MLP_UPDATE_WEIGHTS) animals_net.setActivationFunction(cv2.ml.ANN_MLP_SIGMOID_SYM) # 设置拓扑结构,通过数组来定义各层大小,分别对应输入/隐藏/输出 animals_net.setLayerSizes(np.array([3, 8, 4])) # 指定ann的终止条件 animals_net.setTermCriteria((cv2.TERM_CRITERIA_EPS | cv2.TERM_CRITERIA_COUNT, 10, 1)) """ 输入数组 weight, length, teeth """ """ 输出数组 狗 膺 海豚 龙 dog,eagle,dolphin and dragon """ def dog_sample(): return [randint(5, 20), 1, randint(38, 42)] def dog_class(): return [1, 0, 0, 0] def eagle_sample(): return [randint(3, 13), 3, 0] def eagle_class(): return [0, 1, 0, 0] def dolphin_sample(): return [randint(30, 190), randint(5, 15), randint(80, 100)] def dolphin_class(): return [0, 0, 1, 0] def dragon_sample(): return [randint(1200, 1800), randint(15, 40), randint(160, 180)] def dragon_class(): return [0, 0, 0, 1] """ # 创建四类动物数据,每类5000个样本 SAMPLE = 5000 for x in range(0, SAMPLE): print("samples %d/%d" % (x, SAMPLE)) animals_net.train(np.array([dog_sample()], dtype=np.float32), cv2.ml.ROW_SAMPLE, np.array([dog_class()], dtype=np.float32)) animals_net.train(np.array([eagle_sample()], dtype=np.float32), cv2.ml.ROW_SAMPLE, np.array([eagle_class()], dtype=np.float32)) animals_net.train(np.array([dolphin_sample()], dtype=np.float32), cv2.ml.ROW_SAMPLE, np.array([dolphin_class()], dtype=np.float32)) animals_net.train(np.array([dragon_sample()], dtype=np.float32), cv2.ml.ROW_SAMPLE, np.array([dragon_class()], dtype=np.float32)) print(animals_net.predict(np.array([dog_sample()], dtype=np.float32))) print(animals_net.predict(np.array([eagle_sample()], dtype=np.float32))) print(animals_net.predict(np.array([dolphin_sample()], dtype=np.float32))) print(animals_net.predict(np.array([dragon_sample()], dtype=np.float32))) # 输出结果 # (1.0, array([[ 1.49817729, 1.60551953, -1.56444871, -0.04313202]], dtype=float32)) # (1.0, array([[ 1.49817729, 1.60551953, -1.56444871, -0.04313202]], dtype=float32)) # (1.0, array([[ 1.49817729, 1.60551953, -1.56444871, -0.04313202]], dtype=float32)) # (1.0, array([[ 1.42620921, 1.5461663 , -1.4097836 , 0.07277301]], dtype=float32)) """ # 训练周期 def record(sample, classification): return (np.array([sample], dtype=np.float32), np.array([classification], dtype=np.float32)) records = [] RECORDS = 5000 for x in range(0, RECORDS): records.append(record(dog_sample(), dog_class())) records.append(record(eagle_sample(), eagle_class())) records.append(record(dolphin_sample(), dolphin_class())) records.append(record(dragon_sample(), dragon_class())) EPOCHS = 2 for e in range(0, EPOCHS): print("Epoch %d:" % e) for t, c in records: animals_net.train(t, cv2.ml.ROW_SAMPLE, c) TESTS = 100 dog_results = 0 for x in range(0, TESTS): clas = int(animals_net.predict(np.array([dog_sample()], dtype=np.float32))[0]) print("class: %d" % clas) if (clas) == 0: dog_results += 1 eagle_results = 0 for x in range(0, TESTS): clas = int(animals_net.predict(np.array([eagle_sample()], dtype=np.float32))[0]) print("class: %d" % clas) if (clas) == 1: eagle_results += 1 dolphin_results = 0 for x in range(0, TESTS): clas = int(animals_net.predict(np.array([dolphin_sample()], dtype=np.float32))[0]) print("class: %d" % clas) if (clas) == 2: dolphin_results += 1 dragon_results = 0 for x in range(0, TESTS): clas = int(animals_net.predict(np.array([dragon_sample()], dtype=np.float32))[0]) print("class: %d" % clas) if (clas) == 3: dragon_results += 1 print("Dog accuracy: %f%%" % (dog_results)) print("condor accuracy: %f%%" % (eagle_results)) print("dolphin accuracy: %f%%" % (dolphin_results)) print("dragon accuracy: %f%%" % (dragon_results)) # 输出结果如下: # Dog accuracy: 0.000000% # condor accuracy: 0.000000% # dolphin accuracy: 0.000000% # dragon accuracy: 50.000000%
4 用人工神精网络进行手写数字识别
手写数字数据库,下载地址
http://yann.lecun.com/exdb/mnist
迷你库
#!/usr/bin/env python
# -*- coding: utf-8 -*-
# @Time : 2016/12/17 10:44
# @Author : Retacn
# @Site : opencv ann 手写数字识别
# @File : digits_ann.py
# @Software: PyCharm
__author__ = "retacn"
__copyright__ = "property ofmankind."
__license__ = "CN"
__version__ = "0.0.1"
__maintainer__ = "retacn"
__email__ = "zhenhuayue@sina.com"
__status__ = "Development"
import cv2
import pickle
import numpy as np
import gzip
def load_data():
mnist = gzip.open('./data/mnist.pkl.gz', 'rb')
training_data, classification_data, test_data = pickle.load(mnist,encoding='latin1')
mnist.close()
return (training_data, classification_data, test_data)
def wrap_data():
tr_d, va_d, te_d = load_data()
training_inputs = [np.reshape(x, (784, 1)) for x in tr_d[0]]
training_results = [vectorized_result(y) for y in tr_d[1]]
training_data = zip(training_inputs, training_results)
validation_inputs = [np.reshape(x, (784, 1)) for x in va_d[0]]
validation_data = zip(validation_inputs,va_d[1])
test_inputs = [np.reshape(x, (784, 1)) for x in te_d[0]]
test_data = zip(test_inputs, te_d[1])
return (training_data, validation_data, test_data)
# 给出类标签,创建10个元素的0数组
# 参数j表示要置1的位置
def vectorized_result(j):
e= np.zeros((10, 1))
e[j] = 1.0
return e
# 创建ann
def create_ANN(hidden=20):
ann = cv2.ml.ANN_MLP_create()
#设置各层大小
ann.setLayerSizes(np.array([784, hidden, 10]))
#采用反向传播方式
ann.setTrainMethod(cv2.ml.ANN_MLP_RPROP)
ann.setActivationFunction(cv2.ml.ANN_MLP_SIGMOID_SYM)
#指定ann的终止条件
ann.setTermCriteria((cv2.TERM_CRITERIA_EPS | cv2.TermCriteria_COUNT, 20,1))
return ann
# 训练函数
def train(ann, samples=10000, epochs=1):
tr, val, test = wrap_data()
for x in range(epochs):
counter = 0
for img in tr:
if (counter > samples):
break
if (counter % 1000 == 0):
print("Epoch %d: Trained%d/%d " % (x, counter, samples))
counter += 1
data, digit = img
# ravel()将多维数组拉平为一维
ann.train(np.array([data.ravel()], dtype=np.float32),
cv2.ml.ROW_SAMPLE,
np.array([digit.ravel()],dtype=np.float32))
print('Epoch %d complete' % x)
return ann, test
# 检查神精网络工作
def test(ann, test_data):
sample = np.array(test_data[0][0].ravel(), dtype=np.float32).reshape(28,28)
cv2.imshow("sample", sample)
cv2.waitKey()
print(ann.predict(np.array([test_data[0][0].ravel()],dtype=np.float32)))
def predict(ann, sample):
resized = sample.copy()
rows, cols = resized.shape
if (rows != 28 or cols != 28) and rows * cols > 0:
resized = cv2.resize(resized, (28, 28), interpolation=cv2.INTER_CUBIC)
return ann.predict(np.array([resized.ravel()], dtype=np.float32))
if __name__ == "__main__":
pass
# print(vectorized_result(2))
#!/usr/bin/env python
# -*- coding: utf-8 -*-
# @Time : 2016/12/17 11:35
# @Author : Retacn
# @Site : 识别手写数字图像
# @File : digits_image.py
# @Software: PyCharm
__author__ = "retacn"
__copyright__ = "property ofmankind."
__license__ = "CN"
__version__ = "0.0.1"
__maintainer__ = "retacn"
__email__ = "zhenhuayue@sina.com"
__status__ = "Development"
import cv2
import numpy as np
import Nine.digits_ann as ANN
# 确定矩形是否完全包含在另一个中
def inside(r1, r2):
x1, y1, w1, h1 = r1
x2, y2, w2, h2 = r2
if (x1 > x2) and (y1 > y2) and (x1 + w1 < x2 + w2) and (y1 + h1< y2 + h2):
return True
else:
return False
# 取得数字周围矩形,将其转换为正方形
def wrap_digit(rect):
x, y, w, h = rect
padding = 5
hcenter = x + w / 2
vcenter = y + h / 2
if (h > w):
w = h
x = hcenter - (w / 2)
else:
h = w
y = vcenter - (h / 2)
return (int(x - padding), int(y - padding), int(w + padding), int(h +padding))
# 创建神经网络,中间层为58,训练50000个样本
ann, test_data =ANN.train(ANN.create_ANN(100), 50000,30)
font = cv2.FONT_HERSHEY_SIMPLEX
# 读入图像
PATH = './image/numbers.jpg'
# PATH = './image/MNISTsamples.png'
img = cv2.imread(PATH,cv2.IMREAD_UNCHANGED)
# 更换颜色空间
bw = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
# 高斯模糊
bw = cv2.GaussianBlur(bw, (7, 7), 0)
# 设置阈值
ret, thbw = cv2.threshold(bw, 127, 255,cv2.THRESH_BINARY_INV)
# 腐蚀
thbw = cv2.erode(thbw, np.ones((2, 2),np.uint8), iterations=2)
# 查找轮廓
image, cntrs, hier =cv2.findContours(thbw.copy(), # 源图像
cv2.RETR_TREE, # 模式为查询所有
cv2.CHAIN_APPROX_SIMPLE) # 查询方法
rectangles = []
for c in cntrs:
r= x, y, w, h = cv2.boundingRect(c)
a= cv2.contourArea(c)
b= (img.shape[0] - 3) * (img.shape[1] - 3)
is_inside = False
for q in rectangles:
if inside(r, q):
is_inside = True
break
if not is_inside:
if not a == b:
rectangles.append(r)
# 向预测函数伟递正方形区域
for r in rectangles:
x, y, w, h = wrap_digit(r)
#绘制矩形
cv2.rectangle(img, (x, y), (x + w, y + h), (0,255, 0), 2)
#取得部分图像
roi = thbw[y:y + h, x:x + w]
try:
digit_class = int(ANN.predict(ann, roi.copy())[0])
except:
continue
cv2.putText(img, '%d' % digit_class, (x, y - 1), font, 1, (0, 255, 0))
cv2.imshow("thbw", thbw)
cv2.imshow("contours", img)
cv2.imwrite('./image/sample.jpg', img)
cv2.waitKey()
