机器学习之算法案例手写数字识别

算法案例手写数字识别

MNIST数据集是机器学习领域中非常经典的一个数据集，由60000个
训练样本和10000个测试样本组成，每个样本都是一张28 * 28像素的灰度
手写数字图片。

选择算法，并保存模型

import pickle
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import accuracy_score
from sklearn.ensemble import RandomForestClassifier
import joblib
with open('mnist.pkl','rb') as f:
    train, val ,test = pickle.load(f,encoding='iso-8859-1')

    train_x = train[0]
    train_y = train[1]
    test_x = test[0]
    test_y = test[1]
# lr = LogisticRegression()
# lr.fit(train_x,train_y)
rdt = RandomForestClassifier()
rdt.fit(train_x,train_y)

acc = accuracy_score(rdt.predict(train_x),train_y)
print("训练集上的准确率为：",acc)
acc = accuracy_score(rdt.predict(test_x),test_y)
print("测试集上的准确率为：",acc)

joblib.dump(rdt,'rdt.pkl')

加载模型

给出识别图片

颜色转换

import cv2
img = cv2.imread('1.png')
b = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)
w = 255 - b
cv2.imwrite('9.png',w)

import joblib
import cv2
from sklearn.preprocessing import StandardScaler

rdt = joblib.load('rdt.pkl')

#读取图片
img = cv2.imread('9.png',0)
img = cv2.resize(img,(28,28))
test = img.reshape(1,28*28)

std = StandardScaler()
test = std.fit_transform(test)

pre = rdt.predict(test)
print(pre)

cv2.imshow('img',img)
cv2.waitKey(0)
cv2.destroyAllWindows()

下标为7，查找图片
。。0 。。。1。。。2。。。3。。4。。5。。。6。。7。。。8。。。9