{"id":1170453,"date":"2025-01-15T16:22:43","date_gmt":"2025-01-15T08:22:43","guid":{"rendered":"https:\/\/docs.pingcode.com\/ask\/ask-ask\/1170453.html"},"modified":"2025-01-15T16:22:45","modified_gmt":"2025-01-15T08:22:45","slug":"python%e5%a6%82%e4%bd%95%e5%ae%9e%e7%8e%b0%e5%9b%be%e5%83%8f%e8%af%86%e5%88%ab","status":"publish","type":"post","link":"https:\/\/docs.pingcode.com\/ask\/1170453.html","title":{"rendered":"python\u5982\u4f55\u5b9e\u73b0\u56fe\u50cf\u8bc6\u522b"},"content":{"rendered":"

\"python\u5982\u4f55\u5b9e\u73b0\u56fe\u50cf\u8bc6\u522b\"<\/p>\n

Python\u5b9e\u73b0\u56fe\u50cf\u8bc6\u522b\u7684\u4e3b\u8981\u65b9\u6cd5\u5305\u62ec\uff1a\u4f7f\u7528OpenCV\u5e93\u3001\u4f7f\u7528TensorFlow\u548cKeras\u5e93\u3001\u4f7f\u7528Scikit-learn\u5e93\u3001\u4f7f\u7528PyTorch\u5e93<\/strong>\u3002\u5176\u4e2d\uff0c\u4f7f\u7528TensorFlow\u548cKeras\u5e93\u662f\u76ee\u524d\u6700\u4e3a\u6d41\u884c\u548c\u9ad8\u6548\u7684\u65b9\u6cd5\uff0c\u56e0\u4e3a\u5b83\u4eec\u63d0\u4f9b\u4e86\u5f3a\u5927\u7684\u795e\u7ecf\u7f51\u7edc\u548c\u6df1\u5ea6\u5b66\u4e60\u5de5\u5177\uff0c\u53ef\u4ee5\u5904\u7406\u590d\u6742\u7684\u56fe\u50cf\u8bc6\u522b\u4efb\u52a1\u3002\u4f7f\u7528TensorFlow\u548cKeras\u5e93\u8fdb\u884c\u56fe\u50cf\u8bc6\u522b<\/strong>\uff0c\u6211\u4eec\u9700\u8981\u9996\u5148\u51c6\u5907\u6570\u636e\u96c6\uff0c\u7136\u540e\u6784\u5efa\u548c\u8bad\u7ec3\u6a21\u578b\uff0c\u6700\u540e\u5bf9\u56fe\u50cf\u8fdb\u884c\u9884\u6d4b\u3002\u4e0b\u9762\u5c06\u8be6\u7ec6\u4ecb\u7ecd\u5982\u4f55\u4f7f\u7528TensorFlow\u548cKeras\u5e93\u5b9e\u73b0\u56fe\u50cf\u8bc6\u522b\u3002<\/p>\n<\/p>\n

\u4e00\u3001\u4f7f\u7528OpenCV\u5e93<\/h2>\n<\/p>\n

OpenCV\uff08Open Source Computer Vision Library\uff09\u662f\u4e00\u4e2a\u5f00\u6e90\u8ba1\u7b97\u673a\u89c6\u89c9\u5e93\uff0c\u63d0\u4f9b\u4e86\u8bb8\u591a\u56fe\u50cf\u5904\u7406\u548c\u8ba1\u7b97\u673a\u89c6\u89c9\u7684\u529f\u80fd\u3002\u4f7f\u7528OpenCV\u8fdb\u884c\u56fe\u50cf\u8bc6\u522b\u901a\u5e38\u5305\u62ec\u4ee5\u4e0b\u6b65\u9aa4\uff1a<\/p>\n<\/p>\n

1\u3001\u56fe\u50cf\u9884\u5904\u7406<\/h3>\n<\/p>\n

\u56fe\u50cf\u9884\u5904\u7406\u662f\u56fe\u50cf\u8bc6\u522b\u7684\u7b2c\u4e00\u6b65\uff0c\u5e38\u89c1\u7684\u9884\u5904\u7406\u6b65\u9aa4\u5305\u62ec\uff1a\u7070\u5ea6\u8f6c\u6362\u3001\u4e8c\u503c\u5316\u3001\u6ee4\u6ce2\u3001\u8fb9\u7f18\u68c0\u6d4b\u7b49\u3002\u4f8b\u5982\uff0c\u4f7f\u7528OpenCV\u53ef\u4ee5\u8fdb\u884c\u56fe\u50cf\u7070\u5ea6\u8f6c\u6362\u548c\u4e8c\u503c\u5316\uff1a<\/p>\n<\/p>\n

import cv2<\/p>\n
\u8bfb\u53d6\u56fe\u50cf<\/strong><\/h2>\n
image = cv2.imread('image.jpg')<\/p>\n
\u7070\u5ea6\u8f6c\u6362<\/strong><\/h2>\n
gray_image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)<\/p>\n
\u4e8c\u503c\u5316<\/strong><\/h2>\n
_, binary_image = cv2.threshold(gray_image, 127, 255, cv2.THRESH_BINARY)<\/p>\n
<\/code><\/pre>\n<\/p>\n
2\u3001\u7279\u5f81\u63d0\u53d6<\/h3>\n<\/p>\n
\u5728\u56fe\u50cf\u9884\u5904\u7406\u4e4b\u540e\uff0c\u9700\u8981\u4ece\u56fe\u50cf\u4e2d\u63d0\u53d6\u7279\u5f81\u3002\u5e38\u89c1\u7684\u7279\u5f81\u63d0\u53d6\u65b9\u6cd5\u5305\u62ec\uff1aSIFT\u3001SURF\u3001ORB\u7b49\u3002\u4f8b\u5982\uff0c\u4f7f\u7528ORB\uff08Oriented FAST and Rotated BRIEF\uff09\u8fdb\u884c\u7279\u5f81\u63d0\u53d6\uff1a<\/p>\n<\/p>\n
# \u521b\u5efaORB\u5bf9\u8c61<\/p>\n
orb = cv2.ORB_create()<\/p>\n
\u68c0\u6d4b\u5173\u952e\u70b9\u548c\u8ba1\u7b97\u63cf\u8ff0\u7b26<\/strong><\/h2>\n
keypoints, descriptors = orb.detectAndCompute(gray_image, None)<\/p>\n
\u5728\u56fe\u50cf\u4e2d\u7ed8\u5236\u5173\u952e\u70b9<\/strong><\/h2>\n
image_with_keypoints = cv2.drawKeypoints(image, keypoints, None)<\/p>\n
<\/code><\/pre>\n<\/p>\n
3\u3001\u7279\u5f81\u5339\u914d<\/h3>\n<\/p>\n
\u7279\u5f81\u63d0\u53d6\u4e4b\u540e\uff0c\u9700\u8981\u5c06\u56fe\u50cf\u7684\u7279\u5f81\u4e0e\u5df2\u77e5\u7684\u7279\u5f81\u8fdb\u884c\u5339\u914d\u3002\u5e38\u89c1\u7684\u7279\u5f81\u5339\u914d\u65b9\u6cd5\u5305\u62ec\uff1aBFMatcher\u3001FLANN\u7b49\u3002\u4f8b\u5982\uff0c\u4f7f\u7528BFMatcher\u8fdb\u884c\u7279\u5f81\u5339\u914d\uff1a<\/p>\n<\/p>\n
# \u521b\u5efaBFMatcher\u5bf9\u8c61<\/p>\n
bf = cv2.BFMatcher(cv2.NORM_HAMMING, crossCheck=True)<\/p>\n
\u8fdb\u884c\u7279\u5f81\u5339\u914d<\/strong><\/h2>\n
matches = bf.match(descriptors1, descriptors2)<\/p>\n
\u7ed8\u5236\u5339\u914d\u7ed3\u679c<\/strong><\/h2>\n
matched_image = cv2.drawMatches(image1, keypoints1, image2, keypoints2, matches, None)<\/p>\n
<\/code><\/pre>\n<\/p>\n
\u4e8c\u3001\u4f7f\u7528TensorFlow\u548cKeras\u5e93<\/h2>\n<\/p>\n
TensorFlow\u548cKeras\u662f\u4e24\u4e2a\u5e7f\u6cdb\u4f7f\u7528\u7684\u6df1\u5ea6\u5b66\u4e60\u5e93\uff0c\u53ef\u4ee5\u7528\u4e8e\u6784\u5efa\u548c\u8bad\u7ec3\u590d\u6742\u7684\u795e\u7ecf\u7f51\u7edc\u6a21\u578b\u8fdb\u884c\u56fe\u50cf\u8bc6\u522b\u3002\u4f7f\u7528TensorFlow\u548cKeras\u8fdb\u884c\u56fe\u50cf\u8bc6\u522b\u901a\u5e38\u5305\u62ec\u4ee5\u4e0b\u6b65\u9aa4\uff1a<\/p>\n<\/p>\n
1\u3001\u6570\u636e\u51c6\u5907<\/h3>\n<\/p>\n
\u6570\u636e\u51c6\u5907\u662f\u56fe\u50cf\u8bc6\u522b\u7684\u7b2c\u4e00\u6b65\uff0c\u9700\u8981\u51c6\u5907\u597d\u8bad\u7ec3\u6570\u636e\u96c6\u548c\u6d4b\u8bd5\u6570\u636e\u96c6\u3002\u53ef\u4ee5\u4f7f\u7528Keras\u81ea\u5e26\u7684\u6570\u636e\u96c6\uff0c\u4e5f\u53ef\u4ee5\u4f7f\u7528\u5176\u4ed6\u6570\u636e\u96c6\u3002\u4ee5\u4e0b\u662f\u4f7f\u7528Keras\u81ea\u5e26\u7684CIFAR-10\u6570\u636e\u96c6\u7684\u793a\u4f8b\uff1a<\/p>\n<\/p>\n
import tensorflow as tf<\/p>\n
from tensorflow.keras.datasets import cifar10<\/p>\n
\u52a0\u8f7dCIFAR-10\u6570\u636e\u96c6<\/strong><\/h2>\n
(x_trAI<\/a>n, y_train), (x_test, y_test) = cifar10.load_data()<\/p>\n
\u6570\u636e\u5f52\u4e00\u5316<\/strong><\/h2>\n
x_train, x_test = x_train \/ 255.0, x_test \/ 255.0<\/p>\n
<\/code><\/pre>\n<\/p>\n
2\u3001\u6784\u5efa\u6a21\u578b<\/h3>\n<\/p>\n
\u5728\u6570\u636e\u51c6\u5907\u4e4b\u540e\uff0c\u9700\u8981\u6784\u5efa\u795e\u7ecf\u7f51\u7edc\u6a21\u578b\u3002\u53ef\u4ee5\u4f7f\u7528Keras\u7684Sequential\u6a21\u578b\u6216Functional API\u6784\u5efa\u6a21\u578b\u3002\u4ee5\u4e0b\u662f\u4f7f\u7528Sequential\u6a21\u578b\u6784\u5efa\u4e00\u4e2a\u7b80\u5355\u5377\u79ef\u795e\u7ecf\u7f51\u7edc\u7684\u793a\u4f8b\uff1a<\/p>\n<\/p>\n
from tensorflow.keras.models import Sequential<\/p>\n
from tensorflow.keras.layers import Conv2D, MaxPooling2D, Flatten, Dense<\/p>\n
\u6784\u5efa\u6a21\u578b<\/strong><\/h2>\n
model = Sequential([<\/p>\n
    Conv2D(32, (3, 3), activation='relu', input_shape=(32, 32, 3)),<\/p>\n
    MaxPooling2D((2, 2)),<\/p>\n
    Conv2D(64, (3, 3), activation='relu'),<\/p>\n
    MaxPooling2D((2, 2)),<\/p>\n
    Flatten(),<\/p>\n
    Dense(64, activation='relu'),<\/p>\n
    Dense(10, activation='softmax')<\/p>\n
])<\/p>\n
\u7f16\u8bd1\u6a21\u578b<\/strong><\/h2>\n
model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])<\/p>\n
<\/code><\/pre>\n<\/p>\n
3\u3001\u8bad\u7ec3\u6a21\u578b<\/h3>\n<\/p>\n
\u5728\u6784\u5efa\u597d\u6a21\u578b\u4e4b\u540e\uff0c\u9700\u8981\u4f7f\u7528\u8bad\u7ec3\u6570\u636e\u96c6\u5bf9\u6a21\u578b\u8fdb\u884c\u8bad\u7ec3\u3002\u4ee5\u4e0b\u662f\u8bad\u7ec3\u6a21\u578b\u7684\u793a\u4f8b\uff1a<\/p>\n<\/p>\n
# \u8bad\u7ec3\u6a21\u578b<\/p>\n
model.fit(x_train, y_train, epochs=10, validation_data=(x_test, y_test))<\/p>\n
<\/code><\/pre>\n<\/p>\n
4\u3001\u8bc4\u4f30\u548c\u9884\u6d4b<\/h3>\n<\/p>\n
\u5728\u6a21\u578b\u8bad\u7ec3\u5b8c\u6210\u4e4b\u540e\uff0c\u53ef\u4ee5\u4f7f\u7528\u6d4b\u8bd5\u6570\u636e\u96c6\u5bf9\u6a21\u578b\u8fdb\u884c\u8bc4\u4f30\uff0c\u5e76\u4f7f\u7528\u6a21\u578b\u8fdb\u884c\u56fe\u50cf\u8bc6\u522b\u9884\u6d4b\u3002\u4ee5\u4e0b\u662f\u8bc4\u4f30\u548c\u9884\u6d4b\u7684\u793a\u4f8b\uff1a<\/p>\n<\/p>\n
# \u8bc4\u4f30\u6a21\u578b<\/p>\n
test_loss, test_acc = model.evaluate(x_test, y_test, verbose=2)<\/p>\n
print('\\nTest accuracy:', test_acc)<\/p>\n
\u4f7f\u7528\u6a21\u578b\u8fdb\u884c\u9884\u6d4b<\/strong><\/h2>\n
predictions = model.predict(x_test)<\/p>\n
<\/code><\/pre>\n<\/p>\n
\u4e09\u3001\u4f7f\u7528Scikit-learn\u5e93<\/h2>\n<\/p>\n
Scikit-learn\u662f\u4e00\u4e2a\u5f3a\u5927\u7684\u673a\u5668\u5b66\u4e60<\/a>\u5e93\uff0c\u63d0\u4f9b\u4e86\u8bb8\u591a\u673a\u5668\u5b66\u4e60\u7b97\u6cd5\uff0c\u53ef\u4ee5\u7528\u4e8e\u56fe\u50cf\u8bc6\u522b\u3002\u4f7f\u7528Scikit-learn\u8fdb\u884c\u56fe\u50cf\u8bc6\u522b\u901a\u5e38\u5305\u62ec\u4ee5\u4e0b\u6b65\u9aa4\uff1a<\/p>\n<\/p>\n
1\u3001\u6570\u636e\u51c6\u5907<\/h3>\n<\/p>\n
\u6570\u636e\u51c6\u5907\u662f\u56fe\u50cf\u8bc6\u522b\u7684\u7b2c\u4e00\u6b65\uff0c\u9700\u8981\u51c6\u5907\u597d\u8bad\u7ec3\u6570\u636e\u96c6\u548c\u6d4b\u8bd5\u6570\u636e\u96c6\u3002\u53ef\u4ee5\u4f7f\u7528Scikit-learn\u81ea\u5e26\u7684\u6570\u636e\u96c6\uff0c\u4e5f\u53ef\u4ee5\u4f7f\u7528\u5176\u4ed6\u6570\u636e\u96c6\u3002\u4ee5\u4e0b\u662f\u4f7f\u7528Scikit-learn\u81ea\u5e26\u7684\u624b\u5199\u6570\u5b57\u6570\u636e\u96c6\u7684\u793a\u4f8b\uff1a<\/p>\n<\/p>\n
from sklearn.datasets import load_digits<\/p>\n
\u52a0\u8f7d\u624b\u5199\u6570\u5b57\u6570\u636e\u96c6<\/strong><\/h2>\n
digits = load_digits()<\/p>\n
\u6570\u636e\u5206\u5272<\/strong><\/h2>\n
x_train, x_test, y_train, y_test = train_test_split(digits.data, digits.target, test_size=0.2, random_state=42)<\/p>\n
<\/code><\/pre>\n<\/p>\n
2\u3001\u7279\u5f81\u63d0\u53d6<\/h3>\n<\/p>\n
\u5728\u6570\u636e\u51c6\u5907\u4e4b\u540e\uff0c\u9700\u8981\u5bf9\u56fe\u50cf\u8fdb\u884c\u7279\u5f81\u63d0\u53d6\u3002Scikit-learn\u63d0\u4f9b\u4e86\u8bb8\u591a\u7279\u5f81\u63d0\u53d6\u65b9\u6cd5\uff0c\u4f8b\u5982PCA\uff08\u4e3b\u6210\u5206\u5206\u6790\uff09\u7b49\u3002\u4ee5\u4e0b\u662f\u4f7f\u7528PCA\u8fdb\u884c\u7279\u5f81\u63d0\u53d6\u7684\u793a\u4f8b\uff1a<\/p>\n<\/p>\n
from sklearn.decomposition import PCA<\/p>\n
\u521b\u5efaPCA\u5bf9\u8c61<\/strong><\/h2>\n
pca = PCA(n_components=64)<\/p>\n
\u8fdb\u884c\u7279\u5f81\u63d0\u53d6<\/strong><\/h2>\n
x_train_pca = pca.fit_transform(x_train)<\/p>\n
x_test_pca = pca.transform(x_test)<\/p>\n
<\/code><\/pre>\n<\/p>\n
3\u3001\u8bad\u7ec3\u6a21\u578b<\/h3>\n<\/p>\n
\u5728\u7279\u5f81\u63d0\u53d6\u4e4b\u540e\uff0c\u9700\u8981\u9009\u62e9\u5408\u9002\u7684\u673a\u5668\u5b66\u4e60\u7b97\u6cd5\u5bf9\u6a21\u578b\u8fdb\u884c\u8bad\u7ec3\u3002\u4ee5\u4e0b\u662f\u4f7f\u7528K\u8fd1\u90bb\u7b97\u6cd5\uff08KNN\uff09\u8fdb\u884c\u8bad\u7ec3\u7684\u793a\u4f8b\uff1a<\/p>\n<\/p>\n
from sklearn.neighbors import KNeighborsClassifier<\/p>\n
\u521b\u5efaKNN\u6a21\u578b<\/strong><\/h2>\n
knn = KNeighborsClassifier(n_neighbors=3)<\/p>\n
\u8bad\u7ec3\u6a21\u578b<\/strong><\/h2>\n
knn.fit(x_train_pca, y_train)<\/p>\n
<\/code><\/pre>\n<\/p>\n
4\u3001\u8bc4\u4f30\u548c\u9884\u6d4b<\/h3>\n<\/p>\n
\u5728\u6a21\u578b\u8bad\u7ec3\u5b8c\u6210\u4e4b\u540e\uff0c\u53ef\u4ee5\u4f7f\u7528\u6d4b\u8bd5\u6570\u636e\u96c6\u5bf9\u6a21\u578b\u8fdb\u884c\u8bc4\u4f30\uff0c\u5e76\u4f7f\u7528\u6a21\u578b\u8fdb\u884c\u56fe\u50cf\u8bc6\u522b\u9884\u6d4b\u3002\u4ee5\u4e0b\u662f\u8bc4\u4f30\u548c\u9884\u6d4b\u7684\u793a\u4f8b\uff1a<\/p>\n<\/p>\n
# \u8bc4\u4f30\u6a21\u578b<\/p>\n
score = knn.score(x_test_pca, y_test)<\/p>\n
print('\\nTest accuracy:', score)<\/p>\n
\u4f7f\u7528\u6a21\u578b\u8fdb\u884c\u9884\u6d4b<\/strong><\/h2>\n
predictions = knn.predict(x_test_pca)<\/p>\n
<\/code><\/pre>\n<\/p>\n
\u56db\u3001\u4f7f\u7528PyTorch\u5e93<\/h2>\n<\/p>\n
PyTorch\u662f\u4e00\u4e2a\u5f00\u6e90\u7684\u6df1\u5ea6\u5b66\u4e60\u5e93\uff0c\u63d0\u4f9b\u4e86\u7075\u6d3b\u7684\u795e\u7ecf\u7f51\u7edc\u6784\u5efa\u548c\u8bad\u7ec3\u5de5\u5177\u3002\u4f7f\u7528PyTorch\u8fdb\u884c\u56fe\u50cf\u8bc6\u522b\u901a\u5e38\u5305\u62ec\u4ee5\u4e0b\u6b65\u9aa4\uff1a<\/p>\n<\/p>\n
1\u3001\u6570\u636e\u51c6\u5907<\/h3>\n<\/p>\n
\u6570\u636e\u51c6\u5907\u662f\u56fe\u50cf\u8bc6\u522b\u7684\u7b2c\u4e00\u6b65\uff0c\u9700\u8981\u51c6\u5907\u597d\u8bad\u7ec3\u6570\u636e\u96c6\u548c\u6d4b\u8bd5\u6570\u636e\u96c6\u3002\u53ef\u4ee5\u4f7f\u7528PyTorch\u81ea\u5e26\u7684\u6570\u636e\u96c6\uff0c\u4e5f\u53ef\u4ee5\u4f7f\u7528\u5176\u4ed6\u6570\u636e\u96c6\u3002\u4ee5\u4e0b\u662f\u4f7f\u7528PyTorch\u7684torchvision\u5e93\u52a0\u8f7dCIFAR-10\u6570\u636e\u96c6\u7684\u793a\u4f8b\uff1a<\/p>\n<\/p>\n
import torch<\/p>\n
import torchvision<\/p>\n
import torchvision.transforms as transforms<\/p>\n
\u6570\u636e\u9884\u5904\u7406<\/strong><\/h2>\n
transform = transforms.Compose([<\/p>\n
    transforms.ToTensor(),<\/p>\n
    transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))<\/p>\n
])<\/p>\n
\u52a0\u8f7dCIFAR-10\u6570\u636e\u96c6<\/strong><\/h2>\n
trainset = torchvision.datasets.CIFAR10(root='.\/data', train=True, download=True, transform=transform)<\/p>\n
trainloader = torch.utils.data.DataLoader(trainset, batch_size=4, shuffle=True, num_workers=2)<\/p>\n
testset = torchvision.datasets.CIFAR10(root='.\/data', train=False, download=True, transform=transform)<\/p>\n
testloader = torch.utils.data.DataLoader(testset, batch_size=4, shuffle=False, num_workers=2)<\/p>\n
<\/code><\/pre>\n<\/p>\n
2\u3001\u6784\u5efa\u6a21\u578b<\/h3>\n<\/p>\n
\u5728\u6570\u636e\u51c6\u5907\u4e4b\u540e\uff0c\u9700\u8981\u6784\u5efa\u795e\u7ecf\u7f51\u7edc\u6a21\u578b\u3002\u53ef\u4ee5\u4f7f\u7528PyTorch\u7684nn\u6a21\u5757\u6784\u5efa\u6a21\u578b\u3002\u4ee5\u4e0b\u662f\u6784\u5efa\u4e00\u4e2a\u7b80\u5355\u5377\u79ef\u795e\u7ecf\u7f51\u7edc\u7684\u793a\u4f8b\uff1a<\/p>\n<\/p>\n
import torch.nn as nn<\/p>\n
import torch.nn.functional as F<\/p>\n
class Net(nn.Module):<\/p>\n
    def __init__(self):<\/p>\n
        super(Net, self).__init__()<\/p>\n
        self.conv1 = nn.Conv2d(3, 6, 5)<\/p>\n
        self.pool = nn.MaxPool2d(2, 2)<\/p>\n
        self.conv2 = nn.Conv2d(6, 16, 5)<\/p>\n
        self.fc1 = nn.Linear(16 * 5 * 5, 120)<\/p>\n
        self.fc2 = nn.Linear(120, 84)<\/p>\n
        self.fc3 = nn.Linear(84, 10)<\/p>\n
    def forward(self, x):<\/p>\n
        x = self.pool(F.relu(self.conv1(x)))<\/p>\n
        x = self.pool(F.relu(self.conv2(x)))<\/p>\n
        x = x.view(-1, 16 * 5 * 5)<\/p>\n
        x = F.relu(self.fc1(x))<\/p>\n
        x = F.relu(self.fc2(x))<\/p>\n
        x = self.fc3(x)<\/p>\n
        return x<\/p>\n
net = Net()<\/p>\n
<\/code><\/pre>\n<\/p>\n
3\u3001\u8bad\u7ec3\u6a21\u578b<\/h3>\n<\/p>\n
\u5728\u6784\u5efa\u597d\u6a21\u578b\u4e4b\u540e\uff0c\u9700\u8981\u5b9a\u4e49\u635f\u5931\u51fd\u6570\u548c\u4f18\u5316\u5668\uff0c\u7136\u540e\u4f7f\u7528\u8bad\u7ec3\u6570\u636e\u96c6\u5bf9\u6a21\u578b\u8fdb\u884c\u8bad\u7ec3\u3002\u4ee5\u4e0b\u662f\u8bad\u7ec3\u6a21\u578b\u7684\u793a\u4f8b\uff1a<\/p>\n<\/p>\n
import torch.optim as optim<\/p>\n
\u5b9a\u4e49\u635f\u5931\u51fd\u6570\u548c\u4f18\u5316\u5668<\/strong><\/h2>\n
criterion = nn.CrossEntropyLoss()<\/p>\n
optimizer = optim.SGD(net.parameters(), lr=0.001, momentum=0.9)<\/p>\n
\u8bad\u7ec3\u6a21\u578b<\/strong><\/h2>\n
for epoch in range(2):  # \u591a\u6279\u6b21\u5faa\u73af<\/p>\n
    running_loss = 0.0<\/p>\n
    for i, data in enumerate(trainloader, 0):<\/p>\n
        # \u83b7\u53d6\u8f93\u5165<\/p>\n
        inputs, labels = data<\/p>\n
        # \u5c06\u68af\u5ea6\u7f13\u5b58\u7f6e\u96f6<\/p>\n
        optimizer.zero_grad()<\/p>\n
        # \u524d\u5411\u4f20\u64ad\uff0c\u53cd\u5411\u4f20\u64ad\uff0c\u4f18\u5316<\/p>\n
        outputs = net(inputs)<\/p>\n
        loss = criterion(outputs, labels)<\/p>\n
        loss.backward()<\/p>\n
        optimizer.step()<\/p>\n
        # \u6253\u5370\u7edf\u8ba1\u4fe1\u606f<\/p>\n
        running_loss += loss.item()<\/p>\n
        if i % 2000 == 1999:    # \u6bcf2000\u6279\u6b21\u6253\u5370\u4e00\u6b21<\/p>\n
            print('[%d, %5d] loss: %.3f' % (epoch + 1, i + 1, running_loss \/ 2000))<\/p>\n
            running_loss = 0.0<\/p>\n
print('Finished Training')<\/p>\n
<\/code><\/pre>\n<\/p>\n
4\u3001\u8bc4\u4f30\u548c\u9884\u6d4b<\/h3>\n<\/p>\n
\u5728\u6a21\u578b\u8bad\u7ec3\u5b8c\u6210\u4e4b\u540e\uff0c\u53ef\u4ee5\u4f7f\u7528\u6d4b\u8bd5\u6570\u636e\u96c6\u5bf9\u6a21\u578b\u8fdb\u884c\u8bc4\u4f30\uff0c\u5e76\u4f7f\u7528\u6a21\u578b\u8fdb\u884c\u56fe\u50cf\u8bc6\u522b\u9884\u6d4b\u3002\u4ee5\u4e0b\u662f\u8bc4\u4f30\u548c\u9884\u6d4b\u7684\u793a\u4f8b\uff1a<\/p>\n<\/p>\n
correct = 0<\/p>\n
total = 0<\/p>\n
with torch.no_grad():<\/p>\n
    for data in testloader:<\/p>\n
        images, labels = data<\/p>\n
        outputs = net(images)<\/p>\n
        _, predicted = torch.max(outputs.data, 1)<\/p>\n
        total += labels.size(0)<\/p>\n
        correct += (predicted == labels).sum().item()<\/p>\n
print('Accuracy of the network on the 10000 test images: %d %%' % (100 * correct \/ total))<\/p>\n
<\/code><\/pre>\n<\/p>\n
\u901a\u8fc7\u4e0a\u8ff0\u56db\u79cd\u65b9\u6cd5\uff0c\u53ef\u4ee5\u4f7f\u7528Python\u5b9e\u73b0\u56fe\u50cf\u8bc6\u522b\u3002\u6bcf\u79cd\u65b9\u6cd5\u90fd\u6709\u5176\u4f18\u70b9\u548c\u9002\u7528\u573a\u666f\uff0c\u9009\u62e9\u5408\u9002\u7684\u65b9\u6cd5\u53ef\u4ee5\u63d0\u9ad8\u56fe\u50cf\u8bc6\u522b\u7684\u6548\u679c\u548c\u6548\u7387\u3002\u4f7f\u7528TensorFlow\u548cKeras\u5e93\u8fdb\u884c\u56fe\u50cf\u8bc6\u522b<\/strong>\u662f\u76ee\u524d\u6700\u4e3a\u6d41\u884c\u548c\u9ad8\u6548\u7684\u65b9\u6cd5\uff0c\u56e0\u4e3a\u5b83\u4eec\u63d0\u4f9b\u4e86\u5f3a\u5927\u7684\u795e\u7ecf\u7f51\u7edc\u548c\u6df1\u5ea6\u5b66\u4e60\u5de5\u5177\uff0c\u53ef\u4ee5\u5904\u7406\u590d\u6742\u7684\u56fe\u50cf\u8bc6\u522b\u4efb\u52a1\u3002\u5e0c\u671b\u901a\u8fc7\u672c\u6587\u7684\u4ecb\u7ecd\uff0c\u80fd\u591f\u5e2e\u52a9\u8bfb\u8005\u66f4\u597d\u5730\u7406\u89e3\u548c\u638c\u63e1Python\u56fe\u50cf\u8bc6\u522b\u7684\u5b9e\u73b0\u65b9\u6cd5\u3002<\/p>\n<\/p>\n
\u76f8\u5173\u95ee\u7b54FAQs\uff1a<\/strong><\/h2>\n
 1. \u56fe\u50cf\u8bc6\u522b\u5728Python\u4e2d\u9700\u8981\u54ea\u4e9b\u5e93\uff1f<\/strong>
\u5728Python\u4e2d\u5b9e\u73b0\u56fe\u50cf\u8bc6\u522b\uff0c\u901a\u5e38\u4f7f\u7528\u4e00\u4e9b\u6d41\u884c\u7684\u5e93\uff0c\u6bd4\u5982OpenCV\u3001Pillow\u3001TensorFlow\u548cKeras\u3002OpenCV\u662f\u7528\u4e8e\u8ba1\u7b97\u673a\u89c6\u89c9\u7684\u5f3a\u5927\u5de5\u5177\uff0cPillow\u5219\u662f\u5904\u7406\u56fe\u50cf\u6587\u4ef6\u7684\u57fa\u672c\u5e93\u3002TensorFlow\u548cKeras\u66f4\u9002\u5408\u6df1\u5ea6\u5b66\u4e60\u6a21\u578b\u7684\u6784\u5efa\u548c\u8bad\u7ec3\uff0c\u5c24\u5176\u662f\u5728\u5904\u7406\u590d\u6742\u7684\u56fe\u50cf\u8bc6\u522b\u4efb\u52a1\u65f6\u3002<\/p>\n
2. \u5982\u4f55\u4f7f\u7528Python\u8fdb\u884c\u7b80\u5355\u7684\u56fe\u50cf\u5206\u7c7b\uff1f<\/strong>
\u5b9e\u73b0\u7b80\u5355\u7684\u56fe\u50cf\u5206\u7c7b\u53ef\u4ee5\u4f7f\u7528Keras\u642d\u914d\u9884\u8bad\u7ec3\u7684\u6df1\u5ea6\u5b66\u4e60\u6a21\u578b\uff0c\u4f8b\u5982VGG16\u6216ResNet\u3002\u8fd9\u4e9b\u6a21\u578b\u5df2\u7ecf\u5728\u5927\u89c4\u6a21\u6570\u636e\u96c6\u4e0a\u8fdb\u884c\u4e86\u8bad\u7ec3\uff0c\u7528\u6237\u53ef\u4ee5\u901a\u8fc7\u52a0\u8f7d\u8fd9\u4e9b\u6a21\u578b\u5e76\u5728\u81ea\u5df1\u7684\u6570\u636e\u96c6\u4e0a\u8fdb\u884c\u5fae\u8c03\uff0c\u6765\u5b9e\u73b0\u56fe\u50cf\u5206\u7c7b\u3002\u5177\u4f53\u6b65\u9aa4\u5305\u62ec\u6570\u636e\u9884\u5904\u7406\u3001\u52a0\u8f7d\u6a21\u578b\u3001\u8bad\u7ec3\u548c\u8bc4\u4f30\u6a21\u578b\u7684\u6027\u80fd\u3002<\/p>\n
3. \u56fe\u50cf\u8bc6\u522b\u7684\u5e94\u7528\u573a\u666f\u6709\u54ea\u4e9b\uff1f<\/strong>
\u56fe\u50cf\u8bc6\u522b\u6280\u672f\u5728\u591a\u4e2a\u9886\u57df\u6709\u5e7f\u6cdb\u7684\u5e94\u7528\uff0c\u5982\u5b89\u5168\u76d1\u63a7\u3001\u533b\u7597\u5f71\u50cf\u5206\u6790\u3001\u81ea\u52a8\u9a7e\u9a76\u3001\u793e\u4ea4\u5a92\u4f53\u5185\u5bb9\u7ba1\u7406\u7b49\u3002\u5728\u5546\u4e1a\u9886\u57df\uff0c\u56fe\u50cf\u8bc6\u522b\u53ef\u4ee5\u7528\u4e8e\u5546\u54c1\u5206\u7c7b\u3001\u987e\u5ba2\u884c\u4e3a\u5206\u6790\u7b49\u3002\u5728\u827a\u672f\u548c\u5a31\u4e50\u65b9\u9762\uff0c\u5b83\u4e5f\u88ab\u7528\u4e8e\u56fe\u50cf\u98ce\u683c\u8f6c\u6362\u548c\u751f\u6210\u827a\u672f\u4f5c\u54c1\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"Python\u5b9e\u73b0\u56fe\u50cf\u8bc6\u522b\u7684\u4e3b\u8981\u65b9\u6cd5\u5305\u62ec\uff1a\u4f7f\u7528OpenCV\u5e93\u3001\u4f7f\u7528TensorFlow\u548cKeras\u5e93\u3001\u4f7f\u7528Sci […]","protected":false},"author":3,"featured_media":1170460,"comment_status":"closed","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[37],"tags":[],"acf":[],"_links":{"self":[{"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/posts\/1170453"}],"collection":[{"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/comments?post=1170453"}],"version-history":[{"count":"1","href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/posts\/1170453\/revisions"}],"predecessor-version":[{"id":1170461,"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/posts\/1170453\/revisions\/1170461"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/media\/1170460"}],"wp:attachment":[{"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/media?parent=1170453"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/categories?post=1170453"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/tags?post=1170453"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}