Welcome to our Support Center
Updated

ConvLSTM3D

Description

Setup and add the convolution lstm 3D layer into the model during the definition graph step. Type : polymorphic.

 

Input parameters

 

Graph in : model architecture.

parameters : layer parameters.

cell : cell instance.
return_sequences? : boolean, Whether to return the last output in the output sequence, or the full sequence.
Default value โ€œFalseโ€.
stateful? : boolean, if True, the last state for each sample at index i in a batch will be used as initial state for the sample of index i in the following batch.
Default value โ€œFalseโ€.
lda_coeff : float, defines the coefficient by which the loss derivative will be multiplied before being sent to the previous layer (since during the backward run we go backwards).
Default value โ€œ1โ€.

in/out param :

input_shape : integer array, shape (not including the batch axis). NB : To be used only if it is the first layer of the model.
 output_behavior : enum, setup if the layer is an output layer.
Default โ€œNot Outputโ€โ€‹.

name (optional) : string, name of the layer.

 

Output parameters

 

Graph out : model architecture.

Dimension

Input shape

6D tensor with shape

  • If data_format = โ€˜channels_lastโ€™ : (samples, time, rows, cols, depth, channels).
  • If data_format = โ€˜channels_firstโ€™ : (samples, time, channels, rows, cols, depth).

Output shape

  • if โ€œreturn_sequencesโ€ = True :
    • If data_format = โ€˜channels_lastโ€™ : 6D tensor with shape (samples, timesteps, new_rows, new_cols, new_depth, filters).
    • If data_format = โ€˜channels_firstโ€™ : 6D tensor with shape (samples, timesteps, filters, new_rows, new_cols, new_depth).
  • if โ€œreturn_sequencesโ€ = False :
    • If data_format = โ€˜channels_lastโ€™ : 5D tensor with shape (samples, new_rows, new_cols, new_depth, filters).
    • If data_format = โ€˜channels_firstโ€™ : 5D tensor with shape (samples, filters, new_rows, new_cols, new_depth).

Example

All these exemples are snippets PNG, you can drop these Snippet onto the block diagram and get the depicted code added to your VI (Do not forget to install HAIBAL library to run it).

ConvLSTM3D layer

1 โ€“ Generate a set of data

We generate an array of data of type single and shape [samples = 10, time = 7, channels = 6, rows = 5, cols = 5, depth = 3].

2 โ€“ Define graph

First, we define the first layer of the graph which is an Input layer (explicit input layer method). This layer is setup as an input array shaped [time = 7, channels = 6, rows = 5, cols = 5, depth = 3].
Then we add to the graph the ConvLSTM3D layer.

3 โ€“ Run graph

We call the forward method and retrieve the result with the โ€œPrediction 5Dโ€ method.
This method returns two variables, the first one is the layer information (cluster composed of the layer name, the graph index and the shape of the output layer) and the second one is the prediction with a shape of [samples, filters, new_rows, new_cols, new_depth].
The output dimension depends on the parameters โ€œreturn-sequencesโ€ refer to the chapter โ€œDimensionโ€ of this documentation.

 

ConvLSTM3D layer, batch and dimension

1 โ€“ Generate a set of data

We generate an array of data of type single and shape [number of batch = 9, samples = 10, time = 7, channels = 6, rows = 5, cols = 5, depth = 3].

2 โ€“ Define graph

First, we define the first layer of the graph which is an Input layer (explicit input layer method). This layer is setup as an input array shaped [time = 7, channels = 6, rows = 5, cols = 5, depth = 3].
Then we add to the graph the ConvLSTM3D layer.

3 โ€“ Run graph

We call the forward method and retrieve the result with the โ€œPrediction 5Dโ€ method.
This method returns two variables, the first one is the layer information (cluster composed of the layer name, the graph index and the shape of the output layer) and the second one is the prediction with a shape of [samples, filters, new_rows, new_cols, new_depth].
The output dimension depends on the parameters โ€œreturn-sequencesโ€ refer to the chapter โ€œDimensionโ€ of this documentation.

 

Tags:
Table of Contents