我正在研究用于識別不同蝴蝶的張量流模型。我為此使用了神經網路，我正在從檔案夾中讀取影像，所有資料都在訓練資料集和驗證資料集中拆分，但我想像這樣拆分這些：
(train_images, train_labels), (test_images, test_labels) = datasets.cifar10.load_data()
代替：
train_ds = utils.image_dataset_from_directory(data_dir, validation_split=0.2, subset="training", seed=123, image_size=(img_height, img_width), batch_size=BATCH_SIZE)
val_ds = utils.image_dataset_from_directory(data_dir, validation_split=0.2, subset="validation", seed=123, image_size=(img_height, img_width), batch_size=BATCH_SIZE)
我試過這樣做，但它使我的模型的準確性成為真正的廢話，所以我認為這是不正確的：
train_images = np.concatenate([x for x, y in train_ds], axis=0)
train_labels = np.concatenate([y for x, y in train_ds], axis=0)
test_images = np.concatenate([x for x, y in val_ds], axis=0)
test_labels = np.concatenate([y for x, y in val_ds], axis=0)
我已經嘗試了很多來自 stackoverflow 的方法，但它們也不起作用。
我的型號：
model = tf.keras.Sequential([
# Please reread this link for a better understanding of the data being entered:
#https://www.codespeedy.com/determine-input-shape-in-keras-tensorflow/
layers.Conv2D(32, (3, 3), activation='relu', input_shape=(180, 180, 3)),
layers.MaxPooling2D((2, 2)),
layers.Conv2D(64, (3, 3), activation='relu'),
layers.MaxPooling2D((2, 2), strides=2),
layers.Flatten(),
layers.Dropout(0.2, input_shape=(180, 180, 3)),
layers.Dense(64, activation='relu'),
layers.Dense(5, activation='softmax') # there are 5 classes_names/folders or 5 kinds of butterflies
])
uj5u.com熱心網友回復：
修復了問題：
train_images = np.array([]).reshape((0,180,180,3))
train_labels = np.array([]).reshape(0,)
for x, y in train_ds:
train_images = np.concatenate((train_images, x), axis=0)
train_labels = np.concatenate((train_labels, y), axis=0)
重塑為影像的輸入形狀，0 因為大小是必需的，否則不能添加 x 或 y。
轉載請註明出處，本文鏈接：https://www.uj5u.com/gongcheng/401802.html