Selecting the machine learning algorithm
We already know that we are using Convolution Neural Networks (CNN) for developing this application. You might wonder why we have chosen CNN and not another neural net. You might already know the answer to this question. There are three reasons why we have chosen CNN:
The amount of visual data present nowadays, which is carefully hand-labeled
The affordable computation machines through which GPUs open the door for optimization
The various kinds of architecture of CNN outperforms the other algorithms
Due to these reasons, we have chosen the CNN with SSD. During the development of the baseline model, we will be using MobileNet, which uses CNN with Single Shot Detector (SSD) techniques underneath. So, in this section, we will look at the architecture of the CNN used during the development of the MobileNet. This will help us understand the pre-trained model.
Architecture of the MobileNet SSD model
MobileNet SSD is fast and does the job of object detection...
