Once we run our inference script, it will connect to the camera and localize a box on each frame, as depicted in the following photograph:
Although the model was trained on the location of heads of pets, we can see that it's quite good at localizing the head of a person. This is where you can note the power of generalization of the model.
When you create your own deep learning apps, you might discover that you have a lack of data for your particular application. However, if you relate your specific case to other available datasets, you might be able to find some applicable dataset that, although different, might allow you to successfully train your model.
