- Passing only one channel and removing the other one in the preprocessing. Doing this way it would be more or less similiar to that of grayscale images as we have only m by n matrix.
- Making elements of other two channels zero or 255 (in case of uint8 and 1 in case of double) or some other constant value.
training a CNN with colors
19 visualizaciones (últimos 30 días)
Mostrar comentarios más antiguos
is it possible to train a neural network with single colored images ? ( blue, green, red) ? would it actually learn features?
0 comentarios
Respuestas (1)
Srivardhan Gadila
el 17 de Dic. de 2020
By single colored images I'm assuming that you are referring to single channel (red or green or blue) of RGB (Truecolor) Images. In that case, Yes you can try training your network with single channel.
As per my knowledge there would be two ways of doing so (assume RGB image of size m by n by 3):
After trying out the above, you can make use of activations to check the features your network has learned. You can also refer to Visualize Activations of a Convolutional Neural Network & Deep Learning Tuning and Visualization.
0 comentarios
Ver también
Categorías
Más información sobre Image Data Workflows en Help Center y File Exchange.
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!