def train_autokeras(RESIZE_TRAIN_IMG_DIR, RESIZE_TEST_IMG_DIR, TRAIN_CSV_DIR,
TEST_CSV_DIR, TIME):
# Load images
train_data, train_labels = load_image_dataset(
csv_file_path=TRAIN_CSV_DIR, images_path=RESIZE_TRAIN_IMG_DIR) # 加载数据
test_data, test_labels = load_image_dataset(
csv_file_path=TEST_CSV_DIR, images_path=RESIZE_TEST_IMG_DIR)
train_data = train_data.astype('float32') / 255
test_data = test_data.astype('float32') / 255
clf = ImageClassifier(verbose=True)
clf.fit(train_data, train_labels, time_limit=TIME) # 找最优模型
clf.final_fit(train_data,
train_labels,
test_data,
test_labels,
retrain=True) # 最优模型继续训练
y = clf.evaluate(test_data, test_labels)
print("测试集精确度:", y)
score = clf.evaluate(train_data, train_labels) # score: 0.8139240506329114
print("训练集精确度:", score)
clf.export_keras_model(MODEL_DIR) # 储存
from autokeras.image.image_supervised import ImageClassifier, load_image_dataset
# we created ./all path where we copied all the images
train_path = '../data/all'
train_labels = '../data/labels_train.csv'
x_train, y_train = load_image_dataset(csv_file_path=train_labels,
images_path=train_path)
#x_val, y_val = load_image_dataset(csv_file_path=validation_labels,images_path=validation_path)
clf = ImageClassifier(verbose=True)
# 4 hours search
clf.fit(x_train, y_train, time_limit=4 * 60 * 60)
best_model = clf.export_keras_model()
keras_model = best_model.produce_keras_model('asdf')
keras_model.summary()
# save it
keras_model.save('best.hdf5')
#clf.final_fit(x_train,y_train,x_val,y_val,retrain = True, trainer_args={'max_iter_num':10})
#print(clf.evaluate(x_val,y_val))
# 使用图片识别器
clf = ImageClassifier(verbose=True)
# 给其训练数据和标签，训练的最长时间可以设定，假设为1分钟，autokers会不断找寻最优的网络模型
clf.fit(train_data, train_labels, time_limit=1 * 60)
# 找到最优模型后，再最后进行一次训练和验证
clf.final_fit(train_data,
train_labels,
test_data,
test_labels,
retrain=True)
# 给出评估结果
y = clf.evaluate(test_data, test_labels)
print("evaluate:", y)
# 给一个图片试试预测是否准确
img = load_img(PREDICT_IMG_PATH)
x = img_to_array(img)
x = x.astype('float32') / 255
x = np.reshape(x, (1, IMAGE_SIZE, IMAGE_SIZE, 3))
print("x shape:", x.shape)
# 最后的结果是一个numpy数组，里面是预测值4，意味着是马，说明预测准确
y = clf.predict(x)
print("predict:", y)
# 导出我们生成的模型
clf.export_keras_model(MODEL_DIR)
# 加载模型
model = load_model(MODEL_DIR)
# 将模型导出成可视化图片
plot_model(model, to_file=MODEL_PNG)
x_train.append(img)
# x_train.reshape(256,256,3)
y_train.append(0)
x_train = np.array(x_train)
y_train = np.array(y_train)
for file_name in os.listdir("test/normal"):
img = cv2.imread("test/normal/" + file_name)
x_test.append(img)
# x_train.reshape(256,256,3)
y_test.append(0)
for file_name in os.listdir("test/anomaly"):
img = cv2.imread("test/anomaly/" + file_name)
x_test.append(img)
# x_train.reshape(256,256,3)
y_test.append(0)
x_test = np.array(x_test)
y_test = np.array(y_test)
print(x_train.shape)
print(y_train.shape)
clf = ImageClassifier(verbose=True)
clf.fit(x_train, y_train, time_limit=12 * 60 * 60)
clf.final_fit(x_train, y_train, x_test, y_test, retrain=True)
clf.export_autokeras_model("./autokeras_model.bin") # Auto-Kerasで読み込めるモデルを保存
clf.export_keras_model("./keras_model.bin") # Kerasで読み込めるモデルを保存
acc = clf.evaluate(x_test, y_test)
y_test = []
base_path = "../data-deep-fashion-women/img/"
#Load the data from local file into a dataframe
df = pd.read_csv('../data-deep-fashion-women/img/WOMEN/labels_test.csv')
print(len(df))
for index, row in df.iterrows():
#print(row[0], row[1])
ss = base_path + row[0]
#print(ss)
img = image.load_img(ss, target_size=(224, 224))
img_data = image.img_to_array(img)
image_data_np = np.array(img_data)
x_test.append(image_data_np)
y_test.append(row[1])
from autokeras.image.image_supervised import load_image_dataset
from autokeras.image.image_supervised import ImageClassifier
clf = ImageClassifier(verbose=True)
clf.fit(x_train, y_train, time_limit=10 * 60 * 60) # 10 hours
clf.final_fit(x_train, y_train, x_test, y_test, retrain=True)
y = clf.evaluate(x_test, y_test)
print(y)
clf.export_autokeras_model('./_models/nas_1.h5')
clf.export_keras_model('./_models/nas_2.h5')
clf.load_searcher().load_best_model().produce_keras_model().save(
'./_models/nas_3.h5')