Understanding the concepts of face emotion recognition
We are using Convolutional Neural Network (CNN) to develop the FER application. Earlier, we looked at the basic architecture of CNN. In order to develop FER applications, we will be using the following CNN architecture and optimizer. We are building CNN that is two layers deep. We will be using two fully connected layers and the SoftMax function to categorize the facial emotions.
We will be using several layers made of the convolutional layer, followed by the ReLU (Rectified Linear Unit) layer, followed by the max pooling layer. Refer to the following diagram, which will help you conceptualize the arrangement of the CNN layers. Let's look at the working of CNN. We will cover the following layers:
The convolutional layer
The ReLU layer
The pooling layer
The fully connected layer
The SoftMax layer
Understanding the convolutional layer
In this layer, we will feed our image in the form of pixel values. We are using a sliding window of 3 x 3 dimension...
