def run(method_readFaceRS='lbp'):
# 读取经过上一步特征处理处理的四个输出(lbp/lbp_pca)
X_train, X_test, y_train, y_test = readFaceRS(method_readFaceRS)
# 取age
y_train_age = y_train[['age']]
y_test_age = y_test[['age']]
predict(X_train, X_test, y_train_age, y_test_age)
def run(method_readFaceRS='hog'):
# 读取经过上一步特征处理处理的四个输出(hog/hog_pca)
X_train, X_test, y_train, y_test = readFaceRS(method_readFaceRS)
# 取age
y_train_age = y_train[['age']]
y_test_age = y_test[['age']]
method_name = method_readFaceRS + '_svm'
predict(X_train, X_test, y_train_age, y_test_age, method_name)
def run(method_readFaceRS='resnet'):
# 读取经过上一步特征处理处理的四个输出(ResNet/ResNet_kpca)
X_train, X_test, y_train, y_test = readFaceRS(method_readFaceRS)
# 取age
y_train_age = y_train[['age']]
y_test_age = y_test[['age']]
method_name = method_readFaceRS + '_xgb'
predict(X_train, X_test, y_train_age, y_test_age, method_name)
def run(method_readFaceRS='lbp_pca'):
# 读取经过上一步特征处理处理的四个输出(lbp/lbp_pca)
X_train, X_test, y_train, y_test = readFaceRS(method_readFaceRS)
# 取age
y_train_age = y_train[['age']]
y_test_age = y_test[['age']]
method_name = method_readFaceRS + '_adaboost'
predict(X_train, X_test, y_train_age, y_test_age, method_name)
def run(method_readFaceRS='sift', k=70):
X_train, X_test, y_train, y_test = readFaceRS(method_readFaceRS)
# age
y_train_age = y_train[['age']]
y_test_age = y_test[['age']]
method_name = method_readFaceRS + '_knn'
predict(X_train[:, 1:], X_test[:, 1:], y_train_age, y_test_age, k,
method_name)
def run(method_readFaceRS='densenet',
method_generateUpdateFaceRS='densenet_kpca'):
'''
@param：进行特征提取+降维，第一个参数为特征提取方法，第二个参数为特征提取+降维方法
'''
# 读取已经经过特征处理过的四个文件作为输入
X_train, X_test, y_train, y_test = readFaceRS(method_readFaceRS)
# 特征提取+降维
X_pca_train, X_pca_test = lower_dimen(X_train, X_test)
# 把特征提取+降维和标签输出
generateUpdateFaceRS(X_pca_train, X_pca_test, y_train, y_test,
method_generateUpdateFaceRS)
def run(method_readFaceRS='densenet',
method_generateUpdateFaceRS='densenet_kpca',
n_components=99):
'''
@param：进行densenet+kpca特征处理，第一个参数为densenet，第二个参数为densenet_kpca
'''
# 读取已经经过HOG特征处理过的四个文件作为输入
X_train, X_test, y_train, y_test = readFaceRS(method_readFaceRS)
# HOG+PCA的输出
X_pca_train, X_pca_test = lower_dimen(X_train, X_test, n_components)
# 把HOG+PCA的输出和标签输出到hog_pca目录
generateUpdateFaceRS(X_pca_train, X_pca_test, y_train, y_test,
method_generateUpdateFaceRS)
def run(method_readFaceRS='densetnet'):
X_train, X_test, y_train, y_test = readFaceRS(method_readFaceRS)
# 在标签中取出age列
y_train_age = y_train[['age']]
y_test_age = y_test[['age']]
# 得到one-hot编码
y_train_age_one = pd.get_dummies(y_train_age)
y_test_age_one = pd.get_dummies(y_test_age)
input_size = CNNInputSize(method=method_readFaceRS)
dataset = DataSet(X_train, X_test, y_train_age_one, y_test_age_one,
input_size)
model = build_model(dataset)
train_model(model, dataset)
evaluate_model(model, dataset)
