Amazon Rekognition also offers a powerful tool for performing facial analysis on images. It can predict interesting attributes, such as age and gender, based on looking at the image. It can also detect features such as a smile or whether the person is wearing glasses from this model. Such models can be trained by analyzing a lot of labeled facial images and training an image recognition model to recognize these features. The CNN models that we studied in Chapter 7, Implementing Deep Learning Algorithms, are a good fit for such applications as they can automatically generate feature maps using local receptive fields methodology from the image and detect boxes that contain evidence of these facial features.

The facial analysis demo can be accessed in the same way as the object detection demo. In order to test the model, we picked the painting of Mona...