在使用fit_generator（....）时，我一直想弄清楚我的数据集中的错误是什么
终端返回此错误： ValueError：生成器的输出应该是元组（x，y，sample_weight）或（x，y）。发现：无
我似乎无法满足fit_generator对数据输入和输出的要求，从而导致阵列问题。我看着你是什么元组，发现它只是一个n.size数组。
如果有人愿意看看我的代码并告诉我哪里出错我会很好。谢谢 ：）。与此同时，我会继续尝试解决问题。
import csv
import cv2
import numpy as np
import os
import h5py
import sklearn
from sklearn.utils import shuffle
np.random.seed(0)
def generator(images, measurements, batch_size):
# Generator to process a certain portion of the model at a time
# :param data_dir: The data directory
# :param image_paths: The paths to the images
# :param steer_angles: The steering angles
# :param batch_size: The batch size
# :param is_training: Whether this is training data (True) or validation data (False)
num_samples = len(samples)
while 1: # Loop forever so the generator never terminates
shuffle(samples)
for offset in range(0, num_samples, batch_size):
batch_samples = samples[offset:offset+batch_size]
images = []
measurements = []
for batch_sample in batch_samples:
for i in range(3):
source_path = sample[i]
name = source_path.split('/')[-1]
current_path = '../data7/IMG/' + name
image = cv2.imread(current_path)
images.append(image)
correction = 0.05
# Number was chosen with trail and error, I tried 1 and found the car was jerky.
# I saw David use 0.02 and felt it was too week at times, so I selected 0.05.
measurement = float(sample[3])
measurements.append(measurement)
measurements.append(measurement+correction)
measurements.append(measurement-correction)
# import Images
augmented_images = []
augmented_measurements = []
for image, measurement in zip(images, measurements):
augmented_images.append(image)
augmented_measurements.append(measurement)
flipped_image = cv2.flip(image, 1)
flipped_measurement = float(measurement) * -1.0
augmented_images.append(flipped_image)
augmented_measurements.append(flipped_measurement)
augmented_measurements.append(flipped_measurement+correction)
augmented_measurements.append(flipped_measurement-correction)
# trim image to only see section with road
X_data = np.array(augmented_images)
y_data = np.array(augmented_measurements)
print("X_train ", X_data.shape)
print("y_train ", y_data.shape)
yield sklearn.utils.shuffle(X_data, y_data)
# plot function to be called after running the model
def plot_results(history, num = 0):
#Plot the results
#:param history: The fit model
#:param num: The number for the output file to save to
# Plot training and validation loss
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title('model mean squared error loss')
plt.ylabel('mean squared error loss')
plt.xlabel('epoch')
plt.legend(['training set', 'validation set'], loc='upper right')
plt.savefig('training_validation_loss_plot_' + str(num) + '.jpg')
plt.show()
plt.close()
# import data
samples = []
with open('../data7/driving_log.csv') as csvfile:
reader = csv.reader(csvfile)
for sample in reader:
samples.append(sample)
# split Data into Traning and Validation
test_size = 0.2
random_state = 0
# compile and train the model using the generator function
from sklearn.model_selection import train_test_split
train_samples, validation_samples = train_test_split(samples, test_size=test_size, random_state=random_state)
train_generator = generator(train_samples, batch_size=32)
validation_generator = generator(validation_samples, batch_size=32)
# print('train_generator:', train_generator.shape)
# print('validation_generator:', validation_generator.shape)
# Make lines a numpy array
# # from sklearn.model_selection import train_test_split
# X_train, X_valid, y_train, y_valid = train_test_split(samples, test_size=test_size, random_state=random_state)
# X_train_generator = generator(X_train, batch_size=32)
# X_validation_generator = generator(X_valid, batch_size=32)
# y_train_generator = generator(y_train, batch_size=32)
# y_validation_generator = generator(y_valid, batch_size=32)
# print('X_train shape', X_train.shape)
# print('Y_train shape', y_train.shape)
# print('X_valid shape', X_valid.shape)
# print('y_valid shape', y_valid.shape)
# import infomation from keras for creating the learning model
from keras.models import Sequential
from keras.layers import Flatten, Dense, Lambda, Dropout, Cropping2D
from keras.layers.convolutional import Convolution2D
from keras.layers.pooling import MaxPooling2D
from keras.optimizers import adam
from keras.callbacks import ModelCheckpoint
# model peramiters
keep_prob = 0.5
video_H = 160
viedo_L = 320
layers = 3
crop_H = 25
crop_W = 70
model = Sequential()
# model.add(Lambda(lambda x: x / 255.0 - 0.5, input_shape=(video_H, viedo_L, layers)))
# Preprocess incoming data, centered around zero with small standard deviation
model.add(Lambda(lambda x: x / 255.0 - 1., input_shape=(video_H, viedo_L, layers), output_shape=(video_H, viedo_L, layers)))
model.add(Cropping2D(cropping=((crop_W, crop_H), (0,0))))
model.add(Convolution2D(24,5,5, subsample=(2,2), activation="relu"))
model.add(Convolution2D(36,5,5, subsample=(2,2), activation="relu"))
model.add(Convolution2D(48,5,5, subsample=(2,2), activation="relu"))
model.add(Convolution2D(64,3,3, activation="relu"))
model.add(Convolution2D(64,3,3, activation="relu"))
model.add(Dropout(keep_prob))
model.add(Flatten())
model.add(Dense(100, activation="relu"))
model.add(Dense(50, activation="relu"))
model.add(Dense(10, activation="relu"))
model.add(Dense(1))
model.summary()
#Create a checkpoint of the model
checkpoint = ModelCheckpoint('model-{epoch:03d}.h0',
monitor='val_loss',
verbose=0,
save_best_only=True,
mode='auto')
# set constant numbers for model generation
batch_size = 40
num_epochs = 10
samples_per_epoch = 20000
learn_rate = 0.0001
# create a model and save it
model.compile(loss='mse', optimizer=adam(lr=learn_rate))
#model.fit(images, steering, validation_split=0.2, shuffle=True, nb_epoch=epoch) - Used for Orginal training
history_object = model.fit_generator(train_generator, samples_per_epoch=len(train_samples), validation_data=validation_generator, nb_val_samples=len(validation_samples), callbacks=[checkpoint], nb_epoch=5, verbose=1)
history = model.fit_generator(generator(X_train, y_train, batch_size,),
samples_per_epoch,
num_epochs,
max_q_size=1,
validation_data=generator(X_valid, y_valid, batch_size),
nb_val_samples=len(X_valid),
callbacks=[checkpoint],
verbose=1)
model.save('nvidia05.h5')
plot_results(history_object, 0)
exit()