activation() error for video classification.

Check for incorrect argument data type or missing argument in call to function 'activations'.

Error in LSTMClass (line 38)

sequences{i,1} = activations(netCNN,'somfile.avi',layerName,'OutputAs','rows');

Answer 1

[Harshavardhan]

Hi @Ramesh M

The error occurs because the “activations” function expects image data, such as 4-D arrays or an image datastore. To fix the error, the video frames need to be read and pre-processed first.

% Read video frames
v = VideoReader(‘somefile.avi’);
frames = {};
while hasFrame(v)
    img = readFrame(v);
    frames{end+1} = img;
end

% Convert frames to 4D array
inputImages = cat(4, frames{:}); % [height, width, channels, numFrames]

% Extract features for all frames
features = activations(netCNN, inputImages, layerName, 'OutputAs', 'rows');

For more information on “VideoReader” , “readFrame”, “cat” and “activations”, refer to their respective links below.

https://www.mathworks.com/help/matlab/ref/videoreader.html

https://www.mathworks.com/help/matlab/ref/videoreader.readframe.html

https://www.mathworks.com/help/matlab/ref/double.cat.html

https://www.mathworks.com/help/deep-learning-hdl/ref/dlhdl.workflow.activations.html

Hope this helps.