Hi,
I see that you're looking to train a "BPNN" for an image classification task. The term "BPNN" is primarily found in older literature and typically refers to a standard feedforward network, so I assume that's what you meant.
You can use a feedforward neural network (FFNN) for image classification, but it's generally not ideal, especially for complex image data. Feedforward networks can perform basic image classification tasks, but they face significant limitations compared to convolutional neural networks (CNNs), which are purpose-built for handling image data. This is mainly because they fail to capture spatial information and have been flattened over all the pixels of the image and does have an extremely high parameter count.
Convolutional Neural Networks (CNNs) are preferred for image classification due to their ability to capture spatial hierarchies in images through convolutional layers. These layers apply filters to local regions, enabling the network to detect important features like edges and textures while preserving pixel relationships.
Please go through the following documentation to learn more on how to use CNN for image classification in MATLAB,
Also please go through the following documentation to learn more regarding how to properly post in MATLAB answer to get quick reply,
Hope this helps.
