CNN LSTM video classification

Question

i'm trying to create the vgg+lstm network. in this code seq_len is 1400.

video = Input(shape=(seq_len, 224, 224,3))
cnn_base = VGG16(input_shape=(224, 224, 3), weights = 'imagenet', include_top=False)
cnn_out = GlobalAveragePooling2D()(cnn_base.output)
cnn = Model(cnn_base.input, cnn_out)
cnn.trainable=False

encoded_frames = TimeDistributed(cnn)(video)
encoded_sequence = LSTM(256)(encoded_frames)
hidden_layer = Dense(1024, activation='relu')(encoded_sequence)
outputs = Dense(1)(hidden_layer)
model = Model(video, outputs)
print(model.summary())

history = model.fit(w_train, y_train, epochs=60, batch_size=50, shuffle=True, validation_split=0.2, verbose=10)
print(history.history.keys())

and my error is this:

ValueError: Input 0 is incompatible with layer model_6: expected shape=(None, 1400, 224, 224, 3), found shape=(None, 224, 224, 3)

anybody can help me to solve?

does that help? - https://stackoverflow.com/questions/44583254/valueerror-input-0-is-incompatible-with-layer-lstm-13-expected-ndim-3-found-n — David Thery, Jun 22 '21 at 08:26

score 0 · Answer 1 · answered Sep 12 '21 at 08:16

0

Get rid of seq_len in the input shape, because it generates an 5-dim array with shape=(None, seq_len, 224, 224,3).

answered Sep 12 '21 at 08:16

mlguy

1

CNN LSTM video classification

1 Answers1