What kind of features are extracted by CNN from a face image?

I'm performing Facial emotion recognition using CNN. Im using CNN for feature extraction only and using SVM for classification. My actual question is what kind/type of features are extracted from face images by the CNN. i have used pool5 layer for feature extraction.

although it's just a very huge matrix of numbers but im curious what these numbers means and what are these features

Answer 1

[colordepth]

In CNNs, features extracted from face images evolve from basic edges/textures in early layers to complex patterns (like facial parts or expression cues) in deeper layers. These features are numerical representations the network learns to distinguish emotions, but they are usually not easily interpretable.

To visualize the features of a CNN in MATLAB, you can refer to: https://www.mathworks.com/help/deeplearning/ug/visualize-features-of-a-convolutional-neural-network.html