Is this number of filters, that was missed in explanation?
The number of filters in a Conv layer is always a choice that you make. In other words, what Prof Ng calls a “hyperparameter”. So he has chosen to have 6 filters in the first Conv layer and 16 in the second.
Or if your question is “Yes, I know those are choices, but how did he decide on those choices?” Then the answer is by experience and experimentation. He tried different things and picked the choices that gave good results at reasonable cost.