Face Recognition

Apply face Recognition on the last tweeted image from a specific Twitter handle

Overview

As face recognition is a technique to identify a person from an image or video which has been widely used in various applications, such as security, advertising or social media. The following are some applications:

  • In security applications, it helps in protecting data by identifying the authorized people

  • Advertising:

    • Get closer than ever to your target markets which are called Know Your Customer (KYC).

      • For instance, it makes advertising more targeted by making educated guesses at people’s age and gender.

  • For social media, it automatically recognizes people in photos so it can easily suggest to tagging the photos on social media such as Facebook or Twitter

With the AI capabilities of Baseet.ai, providing a UI and API utilities to accelerate your face recognition application for your business. After using the face recognition training app on your dataset, you can use this app (Face Recognition - Twitter API) to recognize the person by fetching the last tweeted image from a certain Twitter handle.

Architecture

The application built based on the following nodes:

Last Tweet By User:

Fetch the last tweet of a certain Twitter handle including image, text, and hashtags. To fetch tweets from Twitter, you have to provide the credential a created app from Twitter developer website:

  • Consumer key

  • Consumer secret key

  • Access token

  • Access token secret

Detect-face node

Apply Multi-task Cascaded Convolutional Neural Network (MTCNN) on images to detect faces

Import Models node

Face Embedding node

Return embedding vector as a feature using FaceNet model

Predict Label

Apply prediction on the input feature vector to predict the class name

Image Preview Face detection

Draw bounding box around the detected face with the predicted class

Sample Run - APP Editor

Assume you want to fetch the last tweet from this Twitter handle, you have first to fill (tw_case) into the Twitter handle input field on the Last Tweet By User, and your credential of the previous Twitter developer as following:

The credential information are provided by default if you don't have Twitter developer app

  • You have also to select the generated model from training app, in this node:

  • Click on Execute App.

  • Given the most recent tweet of the target user is:

  • Return back to the test app and click on Execute app, after a couple of seconds it shows the successful run in the app log

  • To see the result you have to select the last node (Image preview face detection)

  • Select the output tab in the bottom, and click on the image, it shows a popup of the output image with the detected face along with identifying the person's name

Sample Run - Form

If you want to run any application without going into the app details, Baseet.ai offers this option via Run an Example form in the overview section in the app details page.

For this app, the required input is any Twitter handle (twitter_handle: tw_case) to apply face recognition on the last tweeted image from this account:

Click on Run Example. After the app runs successfully, the output section will be shown as follows:

If the image contains one of the faces that the model is trained on then a bounding box will be drawn on the detected face with the name of the person. To enlarge the image click on the image.

If the last tweet doesn’t contain image, then the following message show:

If the image doesn’t contain people, then this message show (Can't detect face in the tweeted image)