Figure 7.

The CNN model. A network consisting of three main blocks is designed to convert the pressure image into a feature space, which is then fed to one dense layer leading to a logistic regressor for recognition of activities.