我使用Keras（版本 2.2.4）訓(xùn)練了以下模型：# imports ...model = Sequential()model.add(Conv2D(filters=64, kernel_size=5, data_format="channels_last", activation="relu"))model.add(BatchNormalization())model.add(MaxPooling2D(data_format="channels_last"))model.add(Conv2D(filters=32, kernel_size=3, data_format="channels_last", activation="relu"))model.add(BatchNormalization())model.add(MaxPooling2D(data_format="channels_last"))model.add(Flatten(data_format="channels_last"))model.add(Dense(units=256, activation="relu"))model.add(Dense(units=128, activation="relu"))model.add(Dense(units=32, activation="relu"))model.add(Dense(units=8, activation="softmax"))# training ...model.save("model.h5")輸入是 shape 的 28 x 28 灰度圖像(28, 28, 1)。我轉(zhuǎn)換了模型，tensorflowjs_converter現(xiàn)在我想使用TensorFlow.js（版本 1.1.0）將它加載到我的網(wǎng)站中：tf.loadLayersModel('./model/model.json')這會(huì)產(chǎn)生以下錯(cuò)誤：The first layer in a Sequential model must get an `inputShape` or `batchInputShape` argument. at new e (errors.ts:48) at e.add (models.ts:440) at e.fromConfig (models.ts:1020) at vp (generic_utils.ts:277) at nd (serialization.ts:31) at models.ts:299 at common.ts:14 at Object.next (common.ts:14) at o (common.ts:14)如何在無(wú)需重新訓(xùn)練模型的情況下修復(fù)此錯(cuò)誤？
3 回答
開(kāi)滿天機(jī)
TA貢獻(xiàn)1786條經(jīng)驗(yàn) 獲得超13個(gè)贊
最好的方法是改變你的 keras 模型并重新訓(xùn)練。
無(wú)論如何，如果您無(wú)法重新訓(xùn)練您的網(wǎng)絡(luò)，您可以手動(dòng)編輯您的model.json文件。
您需要在model.json文件中找到輸入層并添加
"config": {
...
"batch_input_shape": [
null,
28,
28,
1
]
...
}
暮色呼如
TA貢獻(xiàn)1853條經(jīng)驗(yàn) 獲得超9個(gè)贊
嘗試將您的神經(jīng)網(wǎng)絡(luò)調(diào)整為以下格式：
input_img = Input(batch_shape=(None, 28,28,1))
layer1=Conv2D(filters=64, kernel_size=5, data_format="channels_last", activation="relu")(input_img)
layer2=BatchNormalization()(layer1)
.
.
.
final_layer=Dense(units=8, activation="softmax")(previous_layer)
……等等。在最后：
model = Model(inputs = input_img, outputs = final_layer)
梵蒂岡之花
TA貢獻(xiàn)1900條經(jīng)驗(yàn) 獲得超5個(gè)贊
您必須在 keras 模型的 Conv2D 層中指定輸入形狀。
# imports ...
model = Sequential()
model.add(Conv2D(input_shape=(28, 28, 1), filters=64, kernel_size=5, data_format="channels_last", activation="relu"))
model.add(BatchNormalization())
model.add(MaxPooling2D(data_format="channels_last"))
model.add(Conv2D(filters=32, kernel_size=3, data_format="channels_last", activation="relu"))
model.add(BatchNormalization())
model.add(MaxPooling2D(data_format="channels_last"))
model.add(Flatten(data_format="channels_last"))
model.add(Dense(units=256, activation="relu"))
model.add(Dense(units=128, activation="relu"))
model.add(Dense(units=32, activation="relu"))
model.add(Dense(units=8, activation="softmax"))
# training ...
model.save("model.h5")
添加回答
舉報(bào)
0/150
提交
取消