主要任务是预测输入图像的蒙版。因此,我需要以下数据进行训练:
我还有验证的原始照片。
我准备了某种可以预测输出掩码的神经模型。我准备的keras model configuaration应该具有如下所示的拓扑:
我准备进行培训的代码在那里。
import keras
epochs=100
image_datagen = keras.preprocessing.image.ImageDataGenerator()
mask_datagen = keras.preprocessing.image.ImageDataGenerator()
seed = 1
image_generator = image_datagen.flow_from_directory(
'H:/LABS/ship_detection/test_train/',
color_mode='rgb',batch_size=32,target_size=(768,768),
seed=seed)
mask_generator = mask_datagen.flow_from_directory(
'H:/LABS/ship_detection/test_mask/',
class_mode="categorical",batch_size=32,target_size=(768,768),
seed=seed)
train_generator = zip(image_generator, mask_generator)
model.fit_generator(generator=train_generator,
epochs=epochs,
callbacks=callbacks,steps_per_epoch=1)
但是当我尝试使用生成器进行预测时,我遇到了一个问题:
c:\users\harwister\appdata\local\programs\python\python36\lib\site-packages\keras\engine\training_generator.py in fit_generator(model, generator, steps_per_epoch, epochs, verbose, callbacks, validation_data, validation_steps, class_weight, max_queue_size, workers, use_multiprocessing, shuffle, initial_epoch)
208 batch_size = list(x.values())[0].shape[0]
209 else:
--> 210 batch_size = x.shape[0]
211 batch_logs['batch'] = batch_index
212 batch_logs['size'] = batch_size
AttributeError: 'tuple' object has no attribute 'shape'
我肯定做错了事,但是从这些错误中我什么也听不懂。我在Google中找不到响应的一个简单问题是:如何将两幅图像(输入和输出图像)推入Keras进行训练,训练后如何获得提供输入图像的输出图像?
答案 0 :(得分:0)
由于图像和标签(即蒙版)具有单独的生成器,因此需要将List<String> as = new ArrayList<String>();
HttpTransportProperties.Authenticator basicAuth = new HttpTransportProperties.Authenticator();
as.add(Authenticator.BASIC);
basicAuth.setAuthSchemes(as);
basicAuth.setUsername("ABC");
basicAuth.setPassword("password");
basicAuth.setPreemptiveAuthentication(true);
serviceStub._getServiceClient().getOptions().setProperty(
org.apache.axis2.transport.http.HTTPConstants.AUTHENTICATE,
basicAuthenticator);
参数设置为class_mode
,以防止生成器生成任何标签数组:
None
通过这种方式,image_generator = image_datagen.flow_from_directory(class_mode=None, ...)
mask_generator = mask_datagen.flow_from_directory(class_mode=None, ...)
仅生成输入图像,而image_generator
仅生成蒙版(即真实标签)图像。