Facial Recognition in Live camera

FACIAL RECOGNITION USING PYTHON AND OPENCV WITH LIVE CAMERA FEED

In this article, I will explain the how facial recognition system working and equipments which have needed to develop the project.

Hardware required:

a) PC with Core i5 Processor,16GB RAM,1TB HDD, 2GB GPU

b) HQ IP Camera

c) 10/100 Mbps Network switch

Preparation of Hardware :

The required PC should have a Ubuntu 20.01 operating system and I have installed opencv 4.1.0 working in python3 environment. Opencv installation on ubuntu link is given below.. Link.

( Note : Please follow the steps and should not omit a single command or if error found, rectify the error and then continue). Apart from these, we also need a Numpy library and Haar cascase frontal face classifiers.)

Download haarcascade image classifiers for face recognition.Link.

Installation of software :

First download complete project code and pasted into the folder.Link

To create a complete project on Face Recognition, we must work on 3 very distinct phases:

· Face Captureing ( Capture faces and stored into datasets)

· Train the Dataset(Captured faces are converted in trained file.(i.e .. yml file)

· Run Face Recognition

Please see the demo video on YouTube channel link

The whole installation without error takes minimum of six hours.

In this project results depends on Face captureing so use High quality camera and good lighting. For face recognition, I use 30 samples per face, so the image processing gets slow if you used on lower end PC. The installation of camera should cover the maximum of five persons set in frame. During night time good lighting should be arranged on the captureing area.

Features of the Project:

1) Multiple cameras feed used at same time

2) Sampling faces sets to variable.

3) Open source software used , so patent rights not arrised

4) Small area is required for deployment.

5) It is fully compatible with SBC board also.(Like Raspberry pi4, Jetson Nano .,etc.,).