import osimport numpy as npfrom keras.preprocessing.image import ImageDataGeneratorfrom keras.applications import Xception, VGG16, ResNet50conv_base = VGG16(weights='imagenet',include_top=False,input_shape=(224, 224, 3))base_dir = 'NewDCDatatset'train_dir = os.path.join(base_dir, 'Train')validation_dir = os.path.join(base_dir, 'Validation')test_dir = os.path.join(base_dir, 'Test')datagen = ImageDataGenerator(rescale=1./255)batch_size = 20def extract_features(directory, sample_count):    features = np.zeros(shape=(sample_count, 7 , 7 , 512))    labels = np.zeros(shape=(sample_count))    generator = datagen.flow_from_directory(directory,target_size=(224, 224),batch_size=batch_size,class_mode='categorical')    i = 0    for inputs_batch, labels_batch in generator:        features_batch = conv_base.predict(inputs_batch)        features[i * batch_size : (i + 1) * batch_size] = features_batch        labels[i * batch_size : (i + 1) * batch_size] = labels_batch        i += 1        if i * batch_size >= sample_count:            break    return features, labelstrain_features, train_labels = extract_features(train_dir, 9900*2)validation_features, validation_labels = extract_features(validation_dir, 1300*2)test_features, test_labels = extract_features(test_dir, 2600)train_features = np.reshape(train_features, (9900*2, 7 * 7 * 512))validation_features = np.reshape(validation_features, (2600, 7 * 7 * 512))test_features = np.reshape(test_features, (2600, 7 * 7 * 512))from keras import modelsfrom keras import layersfrom keras import optimizersmodel = models.Sequential()model.add(layers.Dense(256, activation='relu', input_dim=7 * 7 * 512))model.add(layers.Dropout(0.5))model.add(layers.Dense(2, activation='softmax'))model.compile(optimizer='adam',loss='categorical_crossentropy',metrics=['acc'])history = model.fit(train_features, train_labels,epochs=3,batch_size=50,shuffle=True)print(model.evaluate(test_features,test_labels))model.save('TLFACE.h5')

您好，这是我的代码，我遇到了上面的错误。我无法理解应该如何将其形状从 (20) 改为 (20,2)。实际上，我之前在进行二分类任务，现在我想进行多分类任务，但无法解决这个问题。任何帮助都将不胜感激，谢谢。以下是错误的详细信息。

"C:\Program Files\Python37\python.exe" C:/Users/SIBAU/Downloads/face-recognition-with-deep-learning-python-master/extract_text.pyUsing TensorFlow backend.2020-05-28 14:46:37.336572: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'cudart64_101.dll'; dlerror: cudart64_101.dll not found2020-05-28 14:46:37.376451: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.2020-05-28 14:47:08.730218: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'nvcuda.dll'; dlerror: nvcuda.dll not found2020-05-28 14:47:08.730502: E tensorflow/stream_executor/cuda/cuda_driver.cc:351] failed call to cuInit: UNKNOWN ERROR (303)2020-05-28 14:47:08.909845: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:169] retrieving CUDA diagnostic information for host: BILAL-LAPTOP2020-05-28 14:47:08.910341: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:176] hostname: BILAL-LAPTOP2020-05-28 14:47:08.932583: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2Found 80 images belonging to 2 classes.Traceback (most recent call last):  File "C:/Users/SIBAU/Downloads/face-recognition-with-deep-learning-python-master/extract_text.py", line 31, in <module>    train_features, train_labels = extract_features(train_dir, 9900*2)  File "C:/Users/SIBAU/Downloads/face-recognition-with-deep-learning-python-master/extract_text.py", line 25, in extract_features    labels[i * batch_size : (i + 1) * batch_size] = labels_batchValueError: could not broadcast input array from shape (20,2) into shape (20)Process finished with exit code 1

回答：

如果您在进行多类分类（每个输入一个答案，答案可能是n种可能性之一），那么我认为可以通过以下方式解决问题：

generator = datagen.flow_from_directory(directory,target_size=(224, 224),batch_size=batch_size,class_mode='sparse')

根据文档，“categorical”是用于二维独热编码标签的，“binary”是用于一维二进制标签的，“sparse”是用于一维整数标签的。由于您不想要多标签分类，而是多类分类，请使用sparse。sparse的一维是批次维度，而独热编码的二维则是一个维度用于批次，另一个维度用于一组标签。

如果您实际上在进行多标签分类（每个输入有多个输出类，比如在一个图像中标记动物类型和水果类型），那么您可能需要：

labels = np.zeros(shape=(sample_count),2)

这允许每个输入有2个标签（在这种情况下）。