Scribd
Upload
72 views

Face Recognition Web App Tutorial: by Fatih Cagatay Akyon

This document provides an overview of building a face recognition web app. It discusses extracting facial features, using classifiers like SVM and multilayer perceptrons to label faces, and deploying the app on Heroku. The app uses a pretrained model from a paper to embed faces as 128D vectors, then classifies them with Naive Bayes, SVM, or MLP against a Lord of the Rings dataset. The local structure has detection at port 3000 sending regions to recognition at port 4000.

Uploaded by

Mancho Jaramillo
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
72 views

Face Recognition Web App Tutorial: by Fatih Cagatay Akyon

This document provides an overview of building a face recognition web app. It discusses extracting facial features, using classifiers like SVM and multilayer perceptrons to label faces, and deploying the app on Heroku. The app uses a pretrained model from a paper to embed faces as 128D vectors, then classifies them with Naive Bayes, SVM, or MLP against a Lord of the Rings dataset. The local structure has detection at port 3000 sending regions to recognition at port 4000.

Uploaded by

Mancho Jaramillo
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 14

Face Recognition Web App Tutorial

by Fatih Cagatay Akyon

1
Web App Preview
• Live demo URL: https://face-recognition-api-flask.herokuapp.com/

2
Who Am I

Fatih Cagatay Akyon

Machine Learning
Deep Learning
10+ IEEE papers linkedin.com/fcakyon
Computer Vision
5 patents Object Detection
[email protected]
Radar Signal Detection
Modulation Classification

3
Outline
• Facial feature extraction
• SVM, Naïve Bayes, Multi Layer Perceptron
• LOTR Dataset
• Local project structure
• Online project structure
• Pushing our app to Github
• Deploying our app to AWS via Heroku

4
Traditional face recognition
• Extract features

• Compare new face features with db

• Store them in a db

User A User B User C User D

5
Deep learning based facial embedding extraction
Paper url: https://www.cv-foundation.org/openaccess/content_cvpr_2015/app/1A_089.pdf

6
ML based face recognition

ML Based
FaceNet Classifier
Faces Embeddings Face Label
Naïve Bayes
Support Vector Machine
Multi Layer Perceptron
…

128x1
vector

7
Naïve Bayes

8
Support Vector Machine

9
Multi Layer Perceptron

Input 2nd Output


1st Layer Layer Layer
Embeddings Layer

128x1
vector Number 32 32 Number
of neuron neuron of
features classes

10
LOTR Dataset

11
Face detection/ Examples classes

Aragorn Gollum

Frodo Legolas

Gandalf Saruman

12
Independent test set
Local project structure

(http://127.0.0.1:3000/detect)
Sending post request on
(http://127.0.0.1:4000/) /detect method (http://127.0.0.1:3000/)

Face recognition Face detection app


app at port 4000 at port 3000
Receiving detected face cords
and probabilities as json

Applying face recognition


using pretrained model
on face detection regions

14

You might also like