how to design a conv-lstm using the deep network designer

Hi,

i want to construct a deep learning network including the convolutional operation and also need the LSTM, so my network including several convolutionconvolution2dLayers and lstmLayers, but when i tried to construct this network using the deep network designer, it always errors the imageinput,lstm connot exist in the same network, so it let me confused how to construct a network including convolution2dLayer and lstmLayer. The input could be the 2d images or the time series data. So could you help me solve this problem or give me a example or say how to design a conv-lstm using the deep network designer. Thak you!

Best

Answer 1

[Parag]

Hi, you might be facing this issue due to the difference in expected input formats. Convolutional layers are designed for spatial inputs, such as images, which have dimensions like height, width, and channels. On the other hand, LSTM layers are intended for sequential inputs, like time series data, which are organized in time steps.

For more clarity on using a CNN-LSTM network, you can refer this MATLAB documentation:

https://www.mathworks.com/help/deeplearning/ug/sequence-classification-using-cnn-lstm-network.html