当前位置： 首页 > news >正文

如何建设提卡网站学校网站建设策划方案

news 2025/12/26 14:35:00
如何建设提卡网站,学校网站建设策划方案,wordpress去掉域名后缀,马鞍山 做网站原文地址#xff1a; http://ihoge.cn/2018/PCASVM人脸识别.html 加载数据 这里使用的测试数据共包含40位人员照片#xff0c;每个人10张照片。也可登陆http://www.cl.cam.ac.uk/research/dtg/attarchive/facesataglance.html 查看400张照片的缩略图。 import time impo…原文地址 http://ihoge.cn/2018/PCASVM人脸识别.html 加载数据 这里使用的测试数据共包含40位人员照片每个人10张照片。也可登陆http://www.cl.cam.ac.uk/research/dtg/attarchive/facesataglance.html 查看400张照片的缩略图。 import time import logging from sklearn.datasets import fetch_olivetti_faceslogging.basicConfig(level logging.INFO, format%(asctime)s %(message)s) # 这里INFO必须大写data_home code/datasets/ logging.info(开始加载数据) faces fetch_olivetti_faces(data_homedata_home) logging.info(加载完成) 这里做下简单的解释 加载的图片保存在faces变量里sklaern已经把每张照片处理成剪切掉头发部分并且64x64大小且人脸居中显示。在真实生产环境中这一步很重要否则模型将被大量的噪声干扰即照片背景变化的发型等这些特征都应该排除在输入特征之外。最后要成功下载数据集还需要安装Python图片图里工具Pillow否则无法对图片解码。下面输出下数据的概要信息 import numpy as npX faces.data y faces.targettargets np.unique(faces.target) target_names np.array([p%d % t for t in targets]) #给每个人做标签 n_targets target_name.shape[0] n_samples, h, w faces.images.shapeprint(Samples count:{}\nTarget count:{}.format(n_samples, n_targets)) print(Image size:{}x{}\nData shape:{}.format(w, h, X.shape)) Samples count:400 Target count:40 Image size:64x64 Data shape:(400, 4096)由输出可知共有40人照片总量400输入特征(64x644096)个。 为了直观观察数据从每个人物的照片里随机选择一张显示定义下画图工具 其中输入参数images是一个二维数据每一行都是一个图片数据。在加载数据时fech_ollivetti_faces()函数已经自动做了预处理图片的每个像素的RBG值都转换成了[0,1]浮点数。因此画出来的照片也是黑白的。子图片识别领域一般用黑白照片就可以了减少计算量的同时也更加准确。 %matplotlib inline from matplotlib import pyplot as plt def plot_gallery(images, titles, h, w, n_row2, n_col5): # 显示图片阵列plt.figure(figsize(2*n_col, 2.2*n_row),dpi140)plt.subplots_adjust(bottom0, left.01, right.99, top.90, hspace.01)for i in range(n_row * n_col):plt.subplot(n_row, n_col, i1)plt.imshow(images[i].reshape((h,w)), cmapplt.cm.gray)plt.title(titles[i])plt.axis(off)n_row 2 n_col 6sample_images None sample_titles [] for i in range(n_targets):people_images X[yi] # 注意这里传入ipeople_sample_index np.random.randint(0, people_images.shape[0], 1)people_sample_image people_images[people_sample_index, :]if sample_images is not None:sample_images np.concatenate((sample_images, people_sample_image), axis0)else:sample_images people_sample_imagesample_titles.append(target_names[i]) # 这里target_names是在前面生成的标签plot_gallery(sample_images, sample_titles, h, w, n_row, n_col)#代码中X[yi]可以选择除特定的所有照片随机选出来的照片放在sample.images数组对象里最后调用之前定义的函数把照片画出来。 from sklearn.model_selection import train_test_splitX_train, X_test, y_train, y_test train_test_split(X, y, test_size0.2, random_state4) 支持向量机第一次尝试 直接食用支持向量机来实现人脸识别 from time import time from sklearn.svm import SVCt time() clf SVC(class_weightbalanced) clf.fit(X_train, y_train) print(耗时{}秒.format(time() - t)) 耗时1.0220119953155518秒1、接着对人脸数据进行预测使用confusion_matrix查看准确性 from sklearn.metrics import confusion_matrixy_pred clf.predict(X_test) cm confusion_matrix(y_test, y_pred, labelsrange(n_targets)) print(confusion_matrix:\n) # np.set_printoptions(thresholdnp.nan) print(cm[:10]) confusion_matrix:[[0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0][0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0][0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0][0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0][0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0][0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0][0 0 0 0 0 0 0 0 0 0 0 4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0][0 0 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0][0 0 0 0 0 0 0 0 0 0 0 4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0][0 0 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]]上面np.set_printoptions()是为了确保cm完整输出这是因为这个数组是40x40的默认情况下不会全部输出。?? 2、使用classification_report查看准确性 但是很明显输出结果效果很差。 因为confusion_matrix理想的输出是矩阵的对角线上有数组其他地方都为0而且这里很多图片都被预测成索引为12的类别了。我买再来看下classification_report的结果 from sklearn.metrics import classification_reportprint(classification_report(y_test, y_pred, target_names target_names)) #这里y_test和y_pred不要颠倒。 precision recall f1-score supportp0 0.00 0.00 0.00 1p1 0.00 0.00 0.00 3p2 0.00 0.00 0.00 2p3 0.00 0.00 0.00 1p4 0.00 0.00 0.00 1p5 0.00 0.00 0.00 1p6 0.00 0.00 0.00 4p7 0.00 0.00 0.00 2p8 0.00 0.00 0.00 4p9 0.00 0.00 0.00 2p10 0.00 0.00 0.00 1p11 0.00 0.00 0.00 0p12 0.00 0.00 0.00 4p13 0.00 0.00 0.00 4p14 0.00 0.00 0.00 1p15 0.00 0.00 0.00 1p16 0.00 0.00 0.00 3p17 0.00 0.00 0.00 2p18 0.00 0.00 0.00 2p19 0.00 0.00 0.00 2p20 0.00 0.00 0.00 1p21 0.00 0.00 0.00 2p22 0.00 0.00 0.00 3p23 0.00 0.00 0.00 2p24 0.00 0.00 0.00 3p25 0.00 0.00 0.00 3p26 0.00 0.00 0.00 2p27 0.00 0.00 0.00 2p28 0.00 0.00 0.00 0p29 0.00 0.00 0.00 2p30 0.00 0.00 0.00 2p31 0.00 0.00 0.00 3p32 0.00 0.00 0.00 2p33 0.00 0.00 0.00 2p34 0.00 0.00 0.00 0p35 0.00 0.00 0.00 2p36 0.00 0.00 0.00 3p37 0.00 0.00 0.00 1p38 0.00 0.00 0.00 2p39 0.00 0.00 0.00 2avg / total 0.00 0.00 0.00 80/Users/hadoop/anaconda3/lib/python3.6/site-packages/sklearn/metrics/classification.py:1135: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples.precision, predicted, average, warn_for) /Users/hadoop/anaconda3/lib/python3.6/site-packages/sklearn/metrics/classification.py:1137: UndefinedMetricWarning: Recall and F-score are ill-defined and being set to 0.0 in labels with no true samples.recall, true, average, warn_for)结果不出预料果然很差。 这是因为这里把每个像素都作为一个输入特征来处理这样那个数据噪声太严重了模型根本没有办法对训练数据集进行拟合。这里有4096个特征向量可是数据集大小才400个比特征个数少了太多而且分出20%作为测试集。这种情况下根本无法进行准确的训练和预测。 使用PCA来处理数据集 选择kkk值解决上述问题的办法有两种，一个是加大数据样本量（在这里这个不太现实），或者使用PCA给数据降维，值选择前k个最重要的特征。这里我们根据PCA算法来计算失真程度来确定k值。在sklearn里，可以从PCA模型的explained_variance_ratio_变量里获取经PCA处理后的数据还原率。这是一个数组，所有元素求和即可知道选择的k" role="presentation" style="position: relative;">kkk值的数据还原率。随着kkk的增大，数值会无限接近于1。利用这一特征，可以让k" role="presentation" style="position: relative;">kkk取值10300之间每个30取一次样。针对这里的情况选择失真度小于5%即可。 from sklearn.decomposition import PCAprint(Exploring explained variance ratio for dataset ...) candidate_components range(10, 300, 30) explained_ratios [] t time() for c in candidate_components:pca PCA(n_componentsc)X_pca pca.fit_transform(X)explained_ratios.append(np.sum(pca.explained_variance_ratio_)) print(Done in {0:.2f}s.format(time()-t)) Exploring explained variance ratio for dataset ... Done in 2.17splt.figure(figsize(8, 5), dpi100) plt.grid() plt.plot(candidate_components, explained_ratios) plt.xlabel(Number of PCA Components) plt.ylabel(Explained Variance Ratio) plt.title(Explained variance ratio for PCA) plt.yticks(np.arange(0.5, 1.05, .05)) plt.xticks(np.arange(0, 300, 20)); 由上图可知若要保留95%的数据还原率kkk值选择120即可。为了更直观的看不同k" role="presentation" style="position: relative;">kkk值的区别这里画出来体验下 def title_prefix(prefix, title):return {}: {}.format(prefix, title)n_row 1 n_col 5sample_images sample_images[0:5] sample_titles sample_titles[0:5]plotting_images sample_images plotting_titles [title_prefix(orig, t) for t in sample_titles] candidate_components [120, 75, 37, 19, 8] for c in candidate_components:print(Fitting and projecting on PCA(n_components{}) ....format(c))t time()pca PCA(n_componentsc)pca.fit(X)X_sample_pca pca.transform(sample_images)X_sample_inv pca.inverse_transform(X_sample_pca)plotting_images np.concatenate((plotting_images, X_sample_inv), axis0)sample_title_pca [title_prefix({}.format(c), t) for t in sample_titles]plotting_titles np.concatenate((plotting_titles, sample_title_pca), axis0)print(Done in {0:.2f}s.format(time() - t))print(Plotting sample image with different number of PCA conpoments ...) plot_gallery(plotting_images, plotting_titles, h, w,n_row * (len(candidate_components) 1), n_col) Fitting and projecting on PCA(n_components120) ... Done in 0.18s Fitting and projecting on PCA(n_components75) ... Done in 0.14s Fitting and projecting on PCA(n_components37) ... Done in 0.11s Fitting and projecting on PCA(n_components19) ... Done in 0.07s Fitting and projecting on PCA(n_components8) ... Done in 0.06s Plotting sample image with different number of PCA conpoments ...利用GridSearchCV选出最优参数 接下来选择k120k120k=120作为PCA的参数对数据集和测试集进行特征提取然后调用GridSearchCV选出最优参数 n_components 120print(Fitting PCA by using training data ...) t time() pca PCA(n_componentsn_components, svd_solverrandomized, whitenTrue).fit(X_train) print(Done in {0:.2f}s.format(time() - t))print(Projecting input data for PCA ...) t time() X_train_pca pca.transform(X_train) X_test_pca pca.transform(X_test) print(Done in {0:.2f}s.format(time() - t)) Fitting PCA by using training data ... Done in 0.16s Projecting input data for PCA ... Done in 0.01sfrom sklearn.model_selection import GridSearchCVprint(Searching the best parameters for SVC ...) param_grid {C: [1, 5, 10, 50],gamma: [0.0001, 0.0005, 0.001, 0.005, 0.01]} clf GridSearchCV(SVC(kernelrbf, class_weightbalanced), param_grid, verbose2, n_jobs4)# 参数n_jobs4表示启动4个进程 clf clf.fit(X_train_pca, y_train) print(Best parameters found by grid search:) print(clf.best_params_) Searching the best parameters for SVC ... Fitting 3 folds for each of 20 candidates, totalling 60 fits [CV] C1, gamma0.0001 ............................................... [CV] C1, gamma0.0001 ............................................... [CV] C1, gamma0.0001 ............................................... [CV] C1, gamma0.0005 ............................................... [CV] ................................ C1, gamma0.0001, total 0.1s [CV] C1, gamma0.0005 ............................................... [CV] ................................ C1, gamma0.0001, total 0.1s [CV] C1, gamma0.0005 ............................................... [CV] ................................ C1, gamma0.0001, total 0.1s [CV] C1, gamma0.001 ................................................ [CV] ................................ C1, gamma0.0005, total 0.1s [CV] C1, gamma0.001 ................................................ [CV] ................................ C1, gamma0.0005, total 0.1s [CV] C1, gamma0.001 ................................................ [CV] ................................ C1, gamma0.0005, total 0.1s [CV] C1, gamma0.01 ................................................. [CV] ................................. C1, gamma0.001, total 0.1s [CV] C5, gamma0.0001 ............................................... [CV] ................................. C1, gamma0.001, total 0.1s [CV] C5, gamma0.0005 ............................................... [CV] ................................. C1, gamma0.001, total 0.1s [CV] C1, gamma0.005 ................................................ [CV] .................................. C1, gamma0.01, total 0.1s [CV] C1, gamma0.01 ................................................. [CV] ................................ C5, gamma0.0001, total 0.1s [CV] C5, gamma0.0001 ............................................... [CV] ................................ C5, gamma0.0005, total 0.1s [CV] C5, gamma0.001 ................................................ [CV] ................................. C1, gamma0.005, total 0.1s [CV] C1, gamma0.005 ................................................ [CV] .................................. C1, gamma0.01, total 0.1s [CV] C1, gamma0.01 ................................................. [CV] ................................ C5, gamma0.0001, total 0.1s [CV] C5, gamma0.0005 ............................................... [CV] ................................. C5, gamma0.001, total 0.1s [CV] C5, gamma0.001 ................................................ [CV] ................................. C1, gamma0.005, total 0.1s [CV] C1, gamma0.005 ................................................ [CV] ................................ C5, gamma0.0005, total 0.1s [CV] C5, gamma0.0005 ............................................... [CV] .................................. C1, gamma0.01, total 0.1s [CV] C5, gamma0.0001 ............................................... [CV] ................................. C5, gamma0.001, total 0.1s [CV] C5, gamma0.001 ................................................ [CV] ................................. C1, gamma0.005, total 0.1s [CV] ................................ C5, gamma0.0005, total 0.1s [CV] C5, gamma0.005 ................................................ [CV] C5, gamma0.01 ................................................. [CV] ................................ C5, gamma0.0001, total 0.1s [CV] C10, gamma0.0001 .............................................. [CV] ................................. C5, gamma0.001, total 0.1s [CV] C10, gamma0.001 ............................................... [CV] ................................. C5, gamma0.005, total 0.1s [CV] C5, gamma0.005 ................................................ [CV] .................................. C5, gamma0.01, total 0.1s [CV] C5, gamma0.01 ................................................. [CV] ............................... C10, gamma0.0001, total 0.1s [CV] C10, gamma0.0005 .............................................. [CV] ................................ C10, gamma0.001, total 0.1s [CV] C10, gamma0.001 ............................................... [CV] ................................. C5, gamma0.005, total 0.1s [CV] C5, gamma0.005 ................................................ [CV] ............................... C10, gamma0.0005, total 0.1s [CV] C10, gamma0.0005 .............................................. [CV] .................................. C5, gamma0.01, total 0.1s [CV] C10, gamma0.0001 .............................................. [CV] ................................ C10, gamma0.001, total 0.1s [CV] C10, gamma0.001 ............................................... [CV] ................................. C5, gamma0.005, total 0.1s [CV] C5, gamma0.01 ................................................. [CV] ............................... C10, gamma0.0001, total 0.1s [CV] C10, gamma0.0001 .............................................. [CV] ............................... C10, gamma0.0005, total 0.1s [CV] C10, gamma0.0005 .............................................. [CV] ................................ C10, gamma0.001, total 0.1s [CV] C10, gamma0.005 ............................................... [CV] .................................. C5, gamma0.01, total 0.1s [CV] C10, gamma0.005 ............................................... [CV] ............................... C10, gamma0.0001, total 0.1s [CV] C10, gamma0.01 ................................................ [CV] ............................... C10, gamma0.0005, total 0.1s [CV] C50, gamma0.0005 .............................................. [CV] ................................ C10, gamma0.005, total 0.1s [CV] C50, gamma0.001 ............................................... [CV] ................................ C10, gamma0.005, total 0.1s [CV] C10, gamma0.005 ............................................... [CV] ................................. C10, gamma0.01, total 0.1s [CV] C50, gamma0.0001 .............................................. [CV] ............................... C50, gamma0.0005, total 0.1s [CV] C50, gamma0.0005 .............................................. [CV] ................................ C50, gamma0.001, total 0.1s [CV] C50, gamma0.001 ............................................... [CV] ............................... C50, gamma0.0001, total 0.1s [CV] ................................ C10, gamma0.005, total 0.1s [CV] C10, gamma0.01 ................................................ [CV] ............................... C50, gamma0.0005, total 0.1s [CV] C50, gamma0.0001 .............................................. [CV] C50, gamma0.0005 .............................................. [CV] ................................ C50, gamma0.001, total 0.1s [CV] ................................. C10, gamma0.01, total 0.1s [CV] C10, gamma0.01 ................................................ [CV] C50, gamma0.005 ............................................... [CV] ............................... C50, gamma0.0001, total 0.1s [CV] C50, gamma0.0001 .............................................. [CV] ............................... C50, gamma0.0005, total 0.1s [CV] C50, gamma0.001 ............................................... [CV] ................................. C10, gamma0.01, total 0.1s [CV] ................................ C50, gamma0.005, total 0.1s [CV] C50, gamma0.005 ............................................... [CV] C50, gamma0.005 ............................................... [CV] ............................... C50, gamma0.0001, total 0.1s [CV] ................................ C50, gamma0.001, total 0.1s [CV] ................................ C50, gamma0.005, total 0.1s [CV] ................................ C50, gamma0.005, total 0.1s [CV] C50, gamma0.01 ................................................ [CV] ................................. C50, gamma0.01, total 0.0s [CV] C50, gamma0.01 ................................................ [CV] ................................. C50, gamma0.01, total 0.0s [CV] C50, gamma0.01 ................................................ [CV] ................................. C50, gamma0.01, total 0.0s Best parameters found by grid search: {C: 10, gamma: 0.0005}[Parallel(n_jobs4)]: Done 60 out of 60 | elapsed: 1.9s finished测试模型准确性 接着使用这一模型对测试集进行预测并分别使用confusion_matrix和classification_report查看其效果 t time() y_pred clf.best_estimator_.predict(X_test_pca) cm confusion_matrix(y_test, y_pred, labelsrange(n_targets)) print(Done in {0:.2f}.\n.format(time()-t)) print(confusion matrix:) np.set_printoptions(thresholdnp.nan) print(cm[:10]) Done in 0.01.confusion matrix: [[1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0][0 3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0][0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0][0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0][0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0][0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0][0 0 0 0 0 0 3 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0][0 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0][0 0 0 0 0 0 0 0 4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0][0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]]from sklearn.metrics import classification_report print(classification_report(y_test, y_pred, target_namestarget_names)) #这里注意y_test和y_pred位置不要颠倒 precision recall f1-score supportp0 1.00 1.00 1.00 1p1 1.00 1.00 1.00 3p2 1.00 0.50 0.67 2p3 1.00 1.00 1.00 1p4 1.00 1.00 1.00 1p5 1.00 1.00 1.00 1p6 1.00 0.75 0.86 4p7 1.00 1.00 1.00 2p8 1.00 1.00 1.00 4p9 1.00 1.00 1.00 2p10 1.00 1.00 1.00 1p11 1.00 1.00 1.00 4p12 1.00 1.00 1.00 4p13 1.00 1.00 1.00 1p14 1.00 1.00 1.00 1p15 0.75 1.00 0.86 3p16 1.00 1.00 1.00 2p17 1.00 1.00 1.00 2p18 1.00 1.00 1.00 2p19 1.00 1.00 1.00 1p20 1.00 1.00 1.00 2p21 1.00 1.00 1.00 3p22 1.00 1.00 1.00 2p23 1.00 1.00 1.00 3p24 0.75 1.00 0.86 3p25 1.00 1.00 1.00 2p26 1.00 1.00 1.00 2p27 1.00 1.00 1.00 2p28 1.00 1.00 1.00 2p29 1.00 1.00 1.00 3p30 1.00 1.00 1.00 2p31 1.00 1.00 1.00 2p32 1.00 1.00 1.00 2p33 1.00 1.00 1.00 3p34 1.00 1.00 1.00 1p35 1.00 1.00 1.00 2p36 1.00 1.00 1.00 2avg / total 0.98 0.97 0.97 80/Users/hadoop/anaconda3/lib/python3.6/site-packages/sklearn/metrics/classification.py:1428: UserWarning: labels size, 37, does not match size of target_names, 40.format(len(labels), len(target_names))疑问 效果非常乐观但是仍有个问题怎么确定p0p37分别对应的是哪个一个人
http://www.pierceye.com/news/218724/

相关文章：

  • 潜江网站设计商务科技网站建设
  • 合肥企业网站制作网站开发公司简介
  • 德州市建设小学网站文员工作内容
  • 网站域名买卖营销技巧第三季在线观看
  • 西宁市公司网站建设比较网站建设
  • 深圳网站设计公司建设部人才网站
  • 广水住房和城乡建设部网站巅峰网站建设
  • 建设智能家居网站SWOT分析商业网站有哪些
  • 相亲网站男人拉我做外汇h5特效网站欣赏
  • 网站一个多少钱有什么做同城的网站
  • 网站 改版 方案wordpress谷歌字体 4.9
  • 全网引流推广 价格网站百度关键词seo排名优化
  • 信用门户网站建设规范win7 iis 发布asp.net网站
  • 织梦网站添加视频教程如何套用wordpress的源码
  • 网站服务器模式wordpress安装后只有英文版
  • 企业网站建设晋升wordpress 自助建站
  • 网站建设案例建站资源
  • 四大商业网站镇江制作网站的
  • 做网站系统学校宠物交易网站模板
  • 佛山市官网网站建设哪家好瑶海合肥网站建设
  • 微网站app建筑图片大全高清
  • 贵阳金阳网站建设公司网站开发与维护书
  • 做非法网站判什么邢免抵退税在哪个网站做
  • 手机网站如何制作免费网站的内容和功能
  • 宁波网站推广合作商品牌设计包括
  • 辽宁省建设厅官方网站制作一个静态网站的步骤
  • 营销外贸网站建设北京房地产网站建设
  • 青海网站建设公司代理记账如何获取客户
  • 做视频必须知道的一些网站企业搜索
  • 网站定制开发北京淮南网新闻最新消息