The result between two different models, when converting keras lstm to onnx use keras2onnx

[phonPtan]

my model

input_shape = (img_w, img_h, 1)     # (128, 64, 1)

# Make Network
inputs = Input(name='the_input', shape=input_shape, dtype='float32')  # (None, 128, 64, 1)

# Convolution layer (VGG)
inner = Conv2D(64, (3, 3), padding='same', name='conv1', kernel_initializer='he_normal')(inputs)  # (None, 128, 64, 64)
inner = BatchNormalization()(inner)
inner = Activation('relu')(inner)
inner = MaxPooling2D(pool_size=(2, 2), name='max1')(inner)  # (None,64, 32, 64)

inner = Conv2D(128, (3, 3), padding='same', name='conv2', kernel_initializer='he_normal')(inner)  # (None, 64, 32, 128)
inner = BatchNormalization()(inner)
inner = Activation('relu')(inner)
inner = MaxPooling2D(pool_size=(2, 2), name='max2')(inner)  # (None, 32, 16, 128)

inner = Conv2D(256, (3, 3), padding='same', name='conv3', kernel_initializer='he_normal')(inner)  # (None, 32, 16, 256)
inner = BatchNormalization()(inner)
inner = Activation('relu')(inner)
inner = Conv2D(256, (3, 3), padding='same', name='conv4', kernel_initializer='he_normal')(inner)  # (None, 32, 16, 256)
inner = BatchNormalization()(inner)
inner = Activation('relu')(inner)
inner = MaxPooling2D(pool_size=(1, 2), name='max3')(inner)  # (None, 32, 8, 256)

inner = Conv2D(512, (3, 3), padding='same', name='conv5', kernel_initializer='he_normal')(inner)  # (None, 32, 8, 512)
inner = BatchNormalization()(inner)
inner = Activation('relu')(inner)
inner = Conv2D(512, (3, 3), padding='same', name='conv6')(inner)  # (None, 32, 8, 512)
inner = BatchNormalization()(inner)
inner = Activation('relu')(inner)
inner = MaxPooling2D(pool_size=(1, 2), name='max4')(inner)  # (None, 32, 4, 512)

inner = Conv2D(512, (2, 2), padding='same', kernel_initializer='he_normal', name='con7')(inner)  # (None, 32, 4, 512)
inner = BatchNormalization()(inner)
inner = Activation('relu')(inner)
print(inner.shape)

# CNN to RNN
inner = Reshape(target_shape=((27,1024)), name='reshape')(inner)  # (None, 32, 2048)
inner = Dense(64, activation='relu', kernel_initializer='he_normal', name='dense1')(inner)  # (None, 32, 64)

# RNN layer
lstm_1 = LSTM(256, return_sequences=True, kernel_initializer='he_normal', name='lstm1')(inner)  # (None, 32, 512)
lstm_1b = LSTM(256, return_sequences=True, go_backwards=True, kernel_initializer='he_normal', name='lstm1_b')(inner)

lstm1_merged = add([lstm_1, lstm_1b])  # (None, 32, 512) 
lstm1_merged = BatchNormalization()(lstm1_merged)

lstm_2 = LSTM(256, return_sequences=True, kernel_initializer='he_normal', name='lstm2')(lstm1_merged)
lstm_2b = LSTM(256, return_sequences=True, go_backwards=True, kernel_initializer='he_normal', name='lstm2_b')(lstm1_merged)

lstm2_merged = concatenate([lstm_2, lstm_2b])  # (None, 32, 1024)
lstm2_merged = BatchNormalization()(lstm2_merged)

# transforms RNN output to character activations:
inner = Dense(num_classes, kernel_initializer='he_normal',name='dense2')(lstm2_merged) #(None, 32, 63)
y_pred = Activation('softmax', name='softmax')(inner)`

[phonPtan]

https://drive.google.com/drive/folders/1gI-7SvPjRJaLaMQH44FigdqIVtBlMykq?usp=sharing model onnx and h5 file. help me

[jiafatom]

Please use keras2onnx and onnxconverter-common from master branch (not from pypi) right now, we supported lstm go_backwards=True recently and it should work.

[phonPtan]

Thank your