Have fun with VRT

In this section I’d like to show you how to use the geometry information returned by the API. Let’s start with the celebrities.

Recognize Celebrities

Below is a Python script that invokes the API and modifies the image by drawing the bounding box and the landmarks. The answer also contains information about emotions. In this example smile and happiness are detected. Also an unrecognized face is detected.
displaying_celebrities.py

The output is:
Name: Julia Roberts
MatchConfidence: 99.96967315673828
Urls: [‘www.wikidata.org/wiki/Q40523’, ‘www.imdb.com/name/nm0000210’]
Emotion: HAPPY

How to analyze a local image file

It is also possible to analyze a local image file by populating the ImageUrl field according to the data uri scheme.
This feature is available for all the endpoints. Below is an example in python:
base64_example.py

The output of this script is:
Name: Tom Hanks
MatchConfidence: 97.55970001220703
Emotion: HAPPY

Same example using Node.js and Axios:
base64_node.js

Protective Equipment

In the next example, I am interested in identifying who wears a head cover and who is not. In the request, I provide the HEAD_COVER equipment. Below is the Python script that highlights people with head cover in green and those without head cover in red, as shown in the figure.

displaying_ppe.py

Facial analysis

VRT provides highly accurate facial analysis; the response contains information about bounding box, landmarks, age range, gender, emotions, smile, eyeglasses, beard, mustache, etc. For instance, for the following face:

some of the information returned are:
Age range: 18 - 26
Smile: False
Gender: Female
Eyeglasses: False
Beard: False
Mustache: False
Emotion: CALM

Detect Labels

In this section I want to show an example of labeling an image. The following table shows the image and the identified labels:

Image	Labels
	Person, Human, People, Family, Tree, Plant, Clothing, Apparel, Girl, Female, Kid, Child, Hand, Teen

Thanks for your attention and have fun!