def run():
x_train, y_train, x_test, y_test = load_images()
# After loading train and evaluate classifier.
clf = ImageClassifier(verbose=True, augment=False)
clf.fit(x_train, y_train, time_limit=12 * 60 * 60)
clf.final_fit(x_train, y_train, x_test, y_test, retrain=True)
y = clf.evaluate(x_test, y_test)
print(y * 100)
def prepare(self, path, x_train, y_train): global clf # TODO: find a better way to make our own Searcher resume = False if os.path.isdir(self.path): resume = True print(f"Resuming") # TODO: Fix resume after debug clf = self.clf = ImageClassifier(verbose=True, augment=False, path=path, resume=False, search_type=ObservableSearcher)
def train(auto_keras_config_id):
print('auto_keras_config object: ' + str(auto_keras_config_id))
auto_keras_config = AutoKerasConfig.objects.get(id=auto_keras_config_id)
auto_keras_config.status = 'in_progress'
auto_keras_config.save()
# Storing save location for models
try:
dump_file = os.path.join(AUTO_ML_MODELS_PATH, 'auto_keras' + str(datetime.datetime.now()) + '.h5')
print('Files to load: ' + auto_keras_config.training_data_filename)
x = numpy.load(os.path.join(AUTO_ML_DATA_PATH, auto_keras_config.training_data_filename))
y = numpy.load(os.path.join(AUTO_ML_DATA_PATH, auto_keras_config.training_labels_filename))
# x, y = load_ml_data(auto_keras_config.training_data_filename, auto_keras_config.training_labels_filename, False, auto_keras_config.make_one_hot_encoding_task_binary)
# TODO this might not work on low ram machines work, but array has to be 3d
if auto_keras_config.preprocessing_object.input_data_type == 'wav':
array4d = []
i=0
for datapoint in x:
print(i)
x_3d = datapoint.reshape((172,128,10)).transpose()
array4d.append(x_3d)
i+=1
x = numpy.array(array4d)
clf = ImageClassifier(verbose=auto_keras_config.verbose)
start = time.time()
clf.fit(x, y, time_limit=auto_keras_config.time_limit)
end = time.time()
#clf.final_fit(x_train, y_train, x_test, y_test, retrain=True)
#y = clf.evaluate(x_test, y_test)
print("Fitting Success!!!")
# storing the best performer
clf.export_autokeras_model(dump_file)
print('saved!')
#
auto_keras_config.training_time = round(end-start, 2)
auto_keras_config.status = 'success'
auto_keras_config.model_path = dump_file
auto_keras_config.save()
print('Status final ' + auto_keras_config.status)
except Exception as e:
end = time.time()
if 'start' in locals():
print('failed after:' + str(end-start))
auto_keras_config.training_time = round(end-start, 2)
auto_keras_config.status = 'fail'
auto_keras_config.additional_remarks = e
auto_keras_config.save()
def run():
x_train, y_train, x_test, y_test = load_images()
# After loading train and evaluate classifier.
clf = ImageClassifier(verbose=True, augment=False)
clf.fit(x_train, y_train, time_limit=12 * 60 * 60)
clf.final_fit(x_train, y_train, x_test, y_test, retrain=True)
y = clf.evaluate(x_test, y_test)
print(y * 100)
import os
from keras.datasets import mnist
from autokeras import ImageClassifier
from autokeras import pickle_from_file
# Customer temp dir by your own
TEMP_DIR = '/tmp/autokeras_U8KEOQ'
model_file_name = os.path.join(TEMP_DIR, 'test_autokeras_model.pkl')
if __name__ == '__main__':
(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train = x_train.reshape(x_train.shape + (1, ))
x_test = x_test.reshape(x_test.shape + (1, ))
clf = ImageClassifier(verbose=True,
augment=False,
path=TEMP_DIR,
resume=True)
clf.fit(x_train, y_train, time_limit=30 * 60)
clf.final_fit(x_train, y_train, x_test, y_test)
clf.export_autokeras_model(model_file_name)
model = pickle_from_file(model_file_name)
results = model.evaluate(x_test, y_test)
print(results)
import numpy as np
from scipy.stats import norm
import matplotlib.pyplot as plt
from autokeras import ImageClassifier
clf = ImageClassifier(searcher_type='bayesian',
path='/Users/haifeng/cifar10_backup',
resume=True)
searcher = clf.load_searcher()
kernel_matrix = searcher.gpr.kernel_matrix
history = searcher.history
diff = np.zeros(kernel_matrix.shape)
for i, item1 in enumerate(history):
for j, item2 in enumerate(history):
e1 = item1['accuracy']
e2 = item2['accuracy']
diff[i][j] = 1 - abs(e1 - e2)
m1 = np.zeros(kernel_matrix.shape)
m2 = np.zeros(kernel_matrix.shape)
m3 = np.zeros(kernel_matrix.shape)
for i in range(47):
for j in range(47):
m1[i][j] = np.where(
np.array(sorted(kernel_matrix.flatten())) == kernel_matrix[i]
[j])[0][0]
m2[i][j] = np.where(
np.array(sorted(diff.flatten())) == diff[i][j])[0][0]
m1[i][j] = norm.ppf(m1[i][j] / (47.0 * 47.0))
m2[i][j] = norm.ppf(m2[i][j] / (47.0 * 47.0))
m3[i][j] = abs(m1[i][j] - m2[i][j])
# scale the raw pixel intensities to the range [0, 1]
trainX = np.array(trainX, dtype="float16") / 255.0
testX = np.array(testX, dtype="float16") / 255.0
trainY = np.array(trainY)
testY = np.array(testY)
print("[INFO] data matrix: {:.2f}MB".format(trainX.nbytes / (1024 * 1000.0)))
print("[INFO] data shape : {}".format(trainX.shape))
print("[INFO] label shape : {}".format(trainY.shape))
# trainX = trainX.reshape(trainX.shape + (1,))
# testX = testX.reshape(testX.shape + (1,))
print(trainX.shape, trainY.shape, testX.shape, testY.shape)
clf = ImageClassifier(path='autokeras_output/', verbose=True, augment=False)
clf.fit(trainX, trainY, time_limit=12 * 60 * 60)
clf.final_fit(trainX, trainY, testX, testY, retrain=True)
y = clf.evaluate(testX, testY)
print(y * 100)
joblib.dump(clf, 'wfc3_autokeras_model.joblib.save')
'''
from keras.datasets import mnist
from autokeras import ImageClassifier
if __name__ == '__main__':
(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train = x_train.reshape(x_train.shape + (1,))
x_test = x_test.reshape(x_test.shape + (1,))
from keras.datasets import mnist from autokeras import ImageClassifier # loadning mnist from keras (X_train, y_train), (X_test, y_test) = mnist.load_data() # each image has to 3D: 2 coordinates, 1 value (gray scale) X_train = X_train.reshape(X_train.shape + (1,)) X_test = X_test.reshape(X_test.shape + (1,)) clf = ImageClassifier(verbose=True, augment=True) clf.fit(X_train, y_train, time_limit=(1*60*60))
from keras.datasets import mnist
from autokeras import ImageClassifier
import tensorflow
if __name__ == '__main__':
print(tensorflow.__version__)
(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train = x_train.reshape(x_train.shape + (1, ))
x_test = x_test.reshape(x_test.shape + (1, ))
clf = ImageClassifier(verbose=True, augment=False)
clf.fit(x_train, y_train, time_limit=2 * 60)
# clf.final_fit(x_train, y_train, x_test, y_test, retrain=True)
y = clf.evaluate(x_test, y_test)
print(y * 100)
from keras.datasets import mnist
from autokeras import ImageClassifier
from matplotlib import pyplot as plt
if __name__ == '__main__':
(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train = x_train.reshape(x_train.shape + (1, ))
x_test = x_test.reshape(x_test.shape + (1, ))
# Strange!!
# nest line actually calls:
# NetworkModule class from:
# anaconda3/envs/auto-keras/lib/python3.6/site-packages/autokeras/net_module.py
# cos code: "from autokeras import ImageClassifier"
# when you execute it, it will call autokeras from anaconda env's site-packages
# however the grammar checker and the hyper-linker will track it from the local package!!!
# they are different!!!!!
clf = ImageClassifier(verbose=True, augment=False)
clf.fit(x_train, y_train, time_limit=30 * 6000)
clf.final_fit(x_train, y_train, x_test, y_test, retrain=True)
y = clf.evaluate(x_test, y_test)
plt.imshow(x_train[0, :, :, 0], cmap=plt.cm.bone)
print(y * 100)
import os
from keras.datasets import mnist
from autokeras import ImageClassifier
from autokeras.utils import pickle_from_file
# Customer temp dir by your own
TEMP_DIR = '/tmp/autokeras_U8KEOQ'
model_file_name = os.path.join(TEMP_DIR, 'test_autokeras_model.pkl')
if __name__ == '__main__':
(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train = x_train.reshape(x_train.shape + (1,))
x_test = x_test.reshape(x_test.shape + (1,))
clf = ImageClassifier(verbose=True, augment=False, path=TEMP_DIR, resume=True)
clf.fit(x_train, y_train, time_limit=30 * 60)
clf.final_fit(x_train, y_train, x_test, y_test)
clf.export_autokeras_model(model_file_name)
model = pickle_from_file(model_file_name)
results = model.evaluate(x_test, y_test)
print(results)
print(y_train)
for i in range(5):
n = i*20
img = x_train[n,:,:,:].reshape((28,28))
print(y_train[n])
plt.figure()
plt.imshow(img,cmap='gray')
plt.xticks([])
plt.yticks([])
plt.show()
'''
if __name__ == '__main__':
start = time.time()
# 模型构建
model = ImageClassifier(verbose=True)
# 搜索网络模型
model.fit(x_train, y_train, time_limit=1 * 60)
# 验证最优模型
model.final_fit(x_train, y_train, x_train, y_train, retrain=True)
# 给出评估结果
score = model.evaluate(x_train, y_train)
# 识别结果
y_predict = model.predict(x_train)
# y_pred = np.argmax(y_predict,axis=1)
# 精确度
accuracy = accuracy_score(y_train, y_predict)
# 打印出score与accuracy
print('score:', score, ' accuracy:', accuracy)
print(y_predict, y_train)
model_dir = r'./trainer/new_auto_learn_Model.h5'
from keras.datasets import mnist
from autokeras import ImageClassifier
from autokeras.constant import Constant
if __name__ == '__main__':
(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train = x_train.reshape(x_train.shape + (1, ))
x_test = x_test.reshape(x_test.shape + (1, ))
clf = ImageClassifier(verbose=True, augment=False)
clf.fit(x_train, y_train, time_limit=30 * 60)
clf.final_fit(x_train, y_train, x_test, y_test, retrain=True)
y = clf.evaluate(x_test, y_test)
print(y * 100)
clf.export_autokeras_model('my_model-1.h5')
from autokeras import ImageClassifier
from tensorflow.keras.datasets import fashion_mnist as fm
(X_train, y_train), (X_test, y_test) = fm.load_data()
X_train = X_train.astype('float32') / 255.0
X_test = X_test.astype('float32') / 255.0
EPOCHS = 10
classifier = ImageClassifier(seed=9, max_trials=10)
classifier.fit(X_train, y_train, epochs=EPOCHS, verbose=2)
print(classifier.evaluate(X_test, y_test))
import os
# path = '/Users/arron/autokeras/'
# os.chdir(path)
from keras.datasets import mnist
# from autokeras.image.image_supervised import ImageClassifier
from autokeras import ImageClassifier
(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train = x_train.reshape(x_train.shape + (1,))
x_test = x_test.reshape(x_test.shape + (1,))
# 1. 将 verbose 指定为 True 意味着搜索过程将打印在屏幕上供我们查看
clf = ImageClassifier(verbose=True)
# 2. 在 fit 方法中，time_limit 参数是指以秒为单位的搜索时间限制
clf.fit(x_train, y_train, time_limit=12 * 60 * 60)
# 3. final_fit 是模型找到最佳模型架构后进行的最后一次训练。将 retrain 参数指定为 True 意味着将重新初 始化模型的权重
clf.final_fit(x_train, y_train, x_test, y_test, retrain=True)
# 4. 在评估模型在测试集上的效果后，print(y) 将显示模型准确率
y = clf.evaluate(x_test, y_test)
print(y)
# export the model
clf.export_keras_model('my_model.h5')
#
# from keras.models import load_model
# model = load_model('my_model.h5')
#
# from keras.utils import plot_model
# plot_model(model, to_file='my_model.png')
from keras.datasets import mnist
from autokeras import ImageClassifier
if __name__ == '__main__':
(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train = x_train.reshape(x_train.shape + (1,))
x_test = x_test.reshape(x_test.shape + (1,))
clf = ImageClassifier(verbose=True, augment=False)
clf.fit(x_train, y_train, time_limit=12 * 60 * 60)
clf.final_fit(x_train, y_train, x_test, y_test, retrain=True)
y = clf.evaluate(x_test, y_test)
print(y * 100)
"""
#Fixed the error, the excel file contained the DS_Store file
#Loading the dataset
x_train, y_train = load_image_dataset(csv_file_path=train_csv,
images_path=train_dir)
x_test, y_test = load_image_dataset(csv_file_path=test_csv,
images_path=test_dir)
x_train = x_train.reshape(x_train.shape + (1, ))
x_test = x_test.reshape(x_test.shape + (1, ))
clf = ImageClassifier(verbose=True)
#Chose to set the training time to 3600s for testing purposes
clf.fit(x_train, y_train, time_limit=(60 * 60 * 9.5))
clf.final_fit(x_train, y_train, x_test, y_test, retrain=True)
#y gives us the accuracy of our structure
y = clf.evaluate(x_test, y_test)
print(y)
#Verfying if the images are corrupt
for filename in listdir(dir_path):
if filename.endswith('.jpg'):
try:
img = image.open(dir_path + "/" + filename) # open the image file
img.verify() # verify that it is, in fact an image
except (IOError, SyntaxError) as e:
print('Bad file:', filename)
test_label_list.append(plant_folder)
print("[INFO] Image loading completed")
except Exception as e:
print(f"Error : {e}")
y_test_ = label_gener.transform(test_label_list)
y_test = [[i] for i in y_test_]
y_test = np.array(y_test)
x_test = np.array(test_image_list) / 225.0 #scailing 0~1
#______________________MODEL DEF____________________________
clf = ImageClassifier(verbose=True,
path='auto-keras/',
searcher_args={'trainer_args': {
'max_iter_num': 30
}})
clf.fit(x_train, y_train, time_limit=2 * 60 * 60) # hour
print(clf.get_best_model_id())
searcher = clf.load_searcher()
print(searcher.history) #view all model result
# FINAL train with best model
clf.final_fit(x_train,
y_train,
x_test,
y_test,
retrain=False,
trainer_args={'max_iter_num': 3})
def init(path, x_train, y_train):
global clf
# TODO: find a better way to make our own Searcher
clf = ImageClassifier(verbose=True, augment=False, path=path, resume=True)
searcher = ObservableSearcher(clf, x_train.shape[1:], y_train)
import numpy as np
from autokeras import ImageClassifier
from tests.common import clean_dir
if __name__ == '__main__':
x_train = np.random.rand(2, 28, 28, 1)
y_train = np.random.rand(2)
x_test = np.random.rand(1, 28, 28, 1)
y_test = np.random.rand(1)
trainer_args = {'max_iter_num': 0, 'batch_size': 128, 'augment': False}
clf = ImageClassifier(path='tests/resources/temp',
verbose=True,
searcher_args={'trainer_args': trainer_args})
clf.fit(x_train, y_train, time_limit=30)
clf.load_searcher().export_json('./test.json')
clean_dir('tests/resources/temp')
