Automatic Attendance System using Face Recognition ( OpenCV 3.1.0 & Raspberry Pi )

Project Phase

 A Face Recognition system to be used for marking attendance in an organisation for a streamlined and centralized record of Employees or Members.



Phase includes the following stages:
  • A C++ program to detect and store faces. (Detection)
  • A Python Script to maintain and link available faces. (Linking)
  • A C++ program to fetch faces from a camera and compare them with available database. (Recognition)
  • A Python Script to update the record on Google Spreadsheets over a secure wireless connection. (Uploading)
All of this runs on a Raspberry Pi 3.

Phase is and has been my most ambitious project because of the way it works. And also because it is composed of three of my most relished domains: Embedded Linux, Machine Learning and Internet of Things.

Each Functional component took considerable time worth mentioning in this catalog however I also do acknowledge that a lot of hows and whys will still be skipped because they are really large in number.
Last but most important, Kudos to Stack Overflow and every developer that asked relevant questions for this project of mine to be completed even after so many dead ends I have encountered.

Project Phase:

My initial setup was Ubuntu 15.10 on an Intel core i5 laptop. Linux was a choice because I already planned to deploy the final project on a Raspberry Pi. This maintained familiarity with the development platform.

The entire project is too long to discuss every working bit in a single blog post so an extended summary is what this post is.

 What Phase Does:

 It Sees You, Remembers You, Recognizes You and Keeps a note of it on Google Drive !
Pretty Cool when you think about it.
Each task stated above uses a separate program linked together to work seamlessly.

It Sees You:

The Video is captured via the integrated webcam (when developing on ubuntu) and via a USB webcam (when run on Raspbian OS [Raspberry Pi]).This is made easy by always fetching the video from the default connected device.As Pi doesn't come with an inbuilt camera,default device is the USB webcam.Voila!

A C++ program linked with the OpenCV (build from source:make,make install) running a cascade Haar's Frontal Face classifier detects the faces in an image.The task of detection is the following two things:
  • Number of faces in the image
  • Segmenting ,Cropping and Resizing Faces
Detecting One Face and Saving to Database

Detecting One Face and Saving to Database

A Video Documenting the database creation is as shown:






It Remembers You:

 The database of a face is created only when only a single face is detected by the classifier.This ensures that the database of a single individual contains images only of that individual.Before saving the faces are gray scaled and resized to 300 x 300 pixels.

The structure of the folder is as :

.
|-- s1
|   |-- 1.pgm
|   |-- ...
|   |-- 29.pgm
|-- s2
|   |-- 1.pgm
|   |-- ...
|   |-- 29.pgm
...
|-- s40
|   |-- 1.pgm
|   |-- ...
|   |-- 29.pgm

A lot of guidance was received from OpenCV documentation.This includes the above folder structures to store faces.

Root Folder
Database of an Individual

Although the database of images is created successfully,For a program to actually "See" them,It is crucial that every image is properly documented along with the ID of the person they represent.This path creation and Labeling is done by a Python Script.

The Python script creates a record in .csv format which contains :
  • Full Path of the image
  • Label Corresponding to the Image
Since images are stored in a folder named after the id of the person, The name of the folder is infact the Label for our task.

The .csv file created looks like: 



The .CSV file created is used by the next segment to fetch,load and train the Face Recognizer algorithm.

It Recognizes You

 The Task of Face Recognition is done by C++ Program written using OpenCV library.
The Face Recognition module is not native to the official source yet so the additional libraries are built using a new method I came up with as documented here.This method is more reliable than the conventional route.

The program fetches live feed from the default imaging device and processes it frame by frame.

The first task that the program performs is to train its Two classifiers on the training database and labels of images.The Two algorithms used are:


Eigenface is single class specific i.e. It finds the similarities between multiple images of same individual whereas FisherFace finds the differences between different individuals.The Collective and commonly agreed result of both these algorithms trained on the same set of images is used as a confirmation of a prediction.

The Haar's cascade is run to segment the faces which are the evaluated by the two algorithms and predictions are returned by both.The value of prediction is accurate 90% of the trials however it depends on the quality of images in the database.

Video Documenting Face Recognition:





Keeping a note on Google Drive:

 The task of connecting securely to google cloud is done by a python script. It uses the following package to do the task of accessing and updating attendance on google spreadsheet.

  • Oauth2client  (Google Cloud Authentication Client)
  • Gspread   (Google Spreadsheet API client)
  • PyOpenSSL (Python Open SSL package)
The Result of prediction (Roll No. or Unique ID) is given to the Cloud Connect Script as a command line argument. The script fetches the date of current day from the system.These two data elements are enough to mark a student as present.

The Logic here is always a tautology, 
i.e. if a student 'A' arrives before the system ,he is marked as present for the current day.
      if a student 'A' is absent, he never arrives for attendance before the system, hence he is not marked for that day thus stating him absent.

The Python Script connects to a google spreadsheet via valid security credentials and update the attendance onto it.The programming is done in such a way that it handles all the possible scenarios that can arise on the spreadsheet section. Few of the problem -> solution are:
  • Date Row not found -> Create row for Current Date. (When taking attendance on a new day)
  • Roll No not found -> Create column for Roll No. (When database is updated)
  • Date Row found, Roll No column not found -> Add Roll No column and write "Present" in current date row   (Database updated during current day)
  • Date Row not found, Roll No. Column Found -> Add Date Row and write "Present" in current Roll No. column  (Database Intact, Day changed )
The Data for the recognized individual is successfully updated in 3-4 seconds. This is slow compared to execution time of our Recognizer program however keeping in mind all the authorizations and Credential check every time, it for sure is a lead over other unsecured connections.

The Google spreadsheet is edited to give write access to our API token so that there is no conflict of permissions during write task. 

Here is the video of Phase updating the attendance of a detected individual in real time:




The Pi Setup:

The Setup is done with a Dell VGA monitor using an HDMI to VGA converter to connect to Raspberry Pi. Additionally USB Webcam,Keyboard & Mouse are connected via USB port.The webcam lights are kept off because of high current surge of 6 LEDs. They barely make any improvements in lighting conditions anyway.
  • An 8GB Sandisk MicroSD card is loaded with NOOBS and Raspbian OS is installed.
  • OpenCV is built from source using my method for extra modules building as stated here.
  • CodeBlocks is installed from apt-get and code is copied to from the ubuntu system to Pi using a thumb drive.
  • Static path for database storage, database linking,fetching and cloud uploading are set to get around using command line arguments every time.The Detection stage still employs CL arguments to denote the person being databased.
The entire system is enclosed in a box as follows :


The LCD and the glowing Leds are part of a temperature monitoring system. It measures the temperature of the box internals to warn or ward off any heat damage. And that is an entirely different story for a later time.

-----------------------------------------------------------------------------------------------------------------
[ Update 17 November,2018 ] :

The code for a dlib variant of the face detection and recognition project is available for access on my github here : https://github.com/sanjeev309/face-recognition-dlib-tensorflow-knn
You will need to modify the core code to suit your requirement for an attendance system.
Pull requests are welcome.
-----------------------------------------------------------------------------------------------------------------


Success is not final, failure is not fatal: it is the courage to continue that counts

Used :
Code::Blocks IDE
PyCharm IDE
Raspbian OS
Atmel Studio 7.0

Written Monday, November 21, 2016 by Sanjeev Tripathi

 , , , , , ,  26 comments

ThinAVR v0.9: The Slim and Minimalist AVR development Board

A While ago I mentioned the need for a minimalist and cheap embedded development board .You may read my previous post here: Developing Slimmer AVR board

It was that day when I designed the prototype and today, I present to you :




A minimum AVR development board with what you need to get started.

The Initial Silk-less Give-Away model
Smooth Edges and Simplified design

Dimensions as that of a Credit Card!

Ergonomic and Simplified PCB Layout

The Size equivalent to the size of your credit card and powered by an ATmega32 at its heart .It gives ThinAVR the following capabilities:


In addition to the above features that are provided by ATmega32 alone, ThinAVR also comes with:
  • A 16X2 LCD connection port mainly for program debugging operations,connected on Port C
  • Additional Power and Ground Pins for interfacing external circuits
  • External Crystal Oscillator for reliable UART/USART operations.The Internal clock just doesn't make it for UART
  • Mounting holes on all 4 corners for a robust design
  • In-Built ISP Port so that you can always reprogram your board.
  • Push Button Power Switch because Reliability
  • On Board Voltage Regulation for protection from over-voltages
  • Power On Indication LED
  • Ergonomic Dimensions the size of a credit card
  • Sleek Look and Slim design

The board is expected to bridge the gap between a development board and a functional embedded system project by making the platform so affordable that each project has its own ThinAVR.

The Initial batch of ThinAVR has been distributed to aspiring embedded developers for a faster growth curve, proper feedback and a higher adoption rate.

ThinAVR v0.9beta :
-No Silk, Markings with a sharpie
-Female connectors for ports,Male Connector for 16x2 LCD
-12MHz Crystal Oscillator
-Embedded ATmega32
-Power Circuit
-Need ISP Programmer to burn programs.

The Next version therefore will have more or less things.

The Mightiest Tree grew from a tiny seed.
Peace Out.

Written Saturday, October 29, 2016 by Sanjeev Tripathi

 , , ,  No comments

Line Follower using Neural Nets : Part 2 (Designing the Neural Net )

This post is the second in the series of developing a neural net line follower. For the post covering dataset generation, go here

Once data is handy, the next objective in a Neural Net is to plan out its structure.
Neural nets are mostly used for uncertain and non-linear operations as is our task of pattern matching a.k.a Classification.

A neural net has mainly the following layers which map the input to the output
  • Input Layer
  • Hidden Layer(s)
  • Output Layer
The Number of Hidden Layer is a deterministic on the complexity of the task being performed.For example, A deep learning net might contain more hidden layers for a deeper defragmentation of data.However there is a trade off between Depth and Speed of execution, therefore for our case, a single hidden layer would be enough.

The Designed Network looks as depicted here:

8 : 1 mapping through a hidden layer of two nodes (and one bias unit)

Intuition says this will be sufficient for our task as a partition in half can easily tell which side of our sensor strip sees the line.

For details of what a neural network is, I would recommend going through the following site: Neural Networks and Deep Learning

For activation function in our network, I have used tangent sigmoid function:
Tangent Sigmoid Function
For faster development time, the neural net designer tool in MATLAB is used.For the sake of understanding, an algorithmic view is as shown:

x = 1 X 8;                                             {Input from 8 nodes}
W12   =   2 X 8;                                     {Weight Matrix from Input layer to Hidden Layer}
Z1= x*(W12)T = (1X8 * 8X2)= 1X2;   {Mapping from Input layer to Hidden layer}
A1= TanSig(Z1)= 1X2;                         {Activation of Hidden Layer}
W23 = 1 X 2;                                        {Weight Matrix from Hidden layer to Output Layer}
Z2 =A1*(W23)T = (1X2 * 2X1)=1X1; {Mapping from Hidden layer to Output Layer}
A2=TanSig(Z2);                                    {Activation of Output Node}
Y=A2;                                                                    {Output}

 The specifications are set up in the NNFit tool in MATLAB and data obtained from the previous post is used to train the network using Backpropagation Algorithm. After the training completes, the entire process is stored as a script using the prompt window.

By default the script generated contains a lot of redundant information which can be optimised on examination.
The entire script is reduced to the following:

 function [y1] = NNLF(x1)   
 %NNLF neural network simulation function.  
 %  
 % Generated by Neural Network Toolbox function genFunction, 13-Aug-2016 14:40:30.  
 %  
 % [y1] = NNLF(x1) takes these arguments:  
 %  x = Qx8 matrix, input #1  
 % and returns:  
 %  y = Qx1 matrix, output #1  
 % where Q is the number of samples.  
 %#ok<*RPMT0>  
 %(c)Sanjeev Tripathi ( AlphaDataOne.blogspot.in )   
 % ===== NEURAL NETWORK CONSTANTS =====  
 % Layer 1  
 b1 = [-8.8516132798193108e-10;2.1615423176361062];  
 IW1_1 = [-30.312171052276302 -15.230142543653209 -7.5989117276441904 -3.8426480381828529 3.8426480396510354 7.5989117277872076 15.230142543113857 30.312171051352024;1.1787493338995503 1.1684723902794487 -0.30187584946551604 -1.2505266965306716 -0.85655951742083458 -0.61361689937359887 -0.51938433720151178 0.43601182390986715];  
 % Layer 2  
 b2 = -3.5519126834821509e-10;  
 LW2_1 = [1.0000000099939996 2.6043564908190074e-09];  
 % ===== SIMULATION ========  
 Q = size(x1,1); % samples  
 x1 = x1';  
 xp1=2*x1 -1;  
 xp1=cast(xp1,'double');  
 a1 = tansig_apply(b1 + IW1_1*xp1);  
 a2 = repmat(b2,1,Q) + LW2_1*a1;  
 y1=a2;      
 end  
 % Sigmoid Symmetric Transfer Function  
 function a = tansig_apply(n,~)  
 a = 2 ./ (1 + exp(-2*n)) - 1;  
 end

The Neural Net being run uses the pre optimized weights to map the input to the output with an accuracy of 100% (High Variance) as the dataset contained all the possibilities. A demo of our net for different inputs entered manually is as shown:


As stated earlier,
        -1 informs to move left 
        +1 informs to move right 
        ~0 informs to keep moving forward

The network of nodes can compute with accuracy any width,orientation or order of line with absolute accuracy. It can also detect multiple lines simultaneously as the network has been trained for all the possible inputs that it can encounter.

An Implementation on AVR Atmega32 microcontroller to be covered soon.
Stay Tuned for more.

Peace Out

Used:
Matlab 2016a

Written Wednesday, September 07, 2016 by Sanjeev Tripathi

 , , ,  2 comments

Line Follower using Neural Nets : Part 1 (Generating Data Set)

Neural Networks are convenient when mapping a function which behaves non-linearly. However, for Training and Testing purpose , Data is the most important piece of the puzzle.

The Neural Network Line Follower to be designed uses 8 Line sensing elements where each will return
 1 : Line Detected
 0 : Line Not Detected

So the possible dataset is the combination of 8 binary units arranged in different pattern leading to a total of
            28 =  256 Samples of data.

I wrote the following MATLAB script to create the data for training the Neural Network.

 function [bin,out]=DataGen(m)  

 % Function to Generate Training Data set for training neural network of a  
 % line follower  
 %    
 %     [bin,out]=DataGen(number of inputs)  
 %       
 % It is a good practice to have an even number of input units for the  
 % NN Network to be trained  
 n=(2^m)-1;              
 bin=decimalToBinaryVector(0:n);   
 [p,q]=size(bin);             
 out=zeros(p,1);   
 bLeft=bin(:,1:(q/2));  
 bRight=bin(:,((q/2)+1):end);  
 wL=bi2de(bLeft,'left-msb');  
 wR=bi2de(bRight,'right-msb');  
  for i=1:p  
    if wL(i)>wR(i)  
      out(i)=-1;  
    end   
    if wL(i)<wR(i)  
      out(i)=1;  
    end  
    if wL(i)==wR(i)  
      out(i)=0;  
    end    
  end  
 end

Here:
         bin holds the binary input sequence of 8 bits    
         out holds the output corresponding to the input bin

Such that:
          if bin=[1 0 0 0 0 0 0 0]  then out = -1      (Line Sensed on far left : Move Left)
          if bin=[0 0 0 0 0 0 1 0]  then out =  1      (Line sensed on right : Move Right)
          if bin=[0 0 1 0 0 1 0 0]  then out =  0      (Line sensed symetrically : Keep Moving Forward)

 For a Neural Network,  0 is pretty much a perfection so it allocates a really small value (~0) such that it can be considered as zero.

From the script above with 8 as a parameter via the following syntax:
           [bin,out]=DataGen(8);
we get:
           bin = 256 X 8 Input Data Matrix
           out = 256 X 1 Output Data Matrix

In the screenshot of the varibles, Left Segment shows the input from the sensors while we have the expected output in the blue bounded box on right:


These Data elements are ready to be used as Training Parameters for our Neural Network.
Further Development to be covered by subsequent posts.

Update 7/Sep/16 : Read Part II: Designing Neural Network

Peace Out.

Used:
Matlab 2016a

Written Tuesday, August 30, 2016 by Sanjeev Tripathi

 , , ,  No comments

RoboAuto : Arduino Bluetooth Joystick for Android™


HC-05 has an added advantage in embedded systems that pretty much every smartphone is equipped with in built Bluetooth hardware.This opens up a large number of possibilities for access and control of embedded systems.

There are apps that do the task of controlling a system via bluetooth however none sufficed to my requirement. So I developed one,

Presenting  
RoboAuto

An Android Application for communication with HC-05 Bluetooth Module.
RoboAuto is a clean and effective app using which you can control your Robots or Devices with a Simple Joystick!
  
What RoboAuto does:

Program Flow

So, You as an embedded developer have to configure your robot only for the commands received from the app!

Here are few screenshots:






RoboAuto connects to the Bluetooth Module using a Wireless Serial Link. Android already provides support for this in their native API.

After a successful connection, characters are sent to the module on actuation of joystick ,buttons or check boxes which can be read and executed by Arduino™ or other microcontrollers.

Special Thanks to zerokol for his JoystickView Library.
The Joystick sends the following :

Forward: F
Backward: B
Left : L
Right : R

Additionally I have assigned intermediate directions , for example, FL for Front-Left, BR for Backward-Right and so on.
No Changes are to be required on the Embedded side as the UART only parses a single character at a time, giving us simultaneous (in theory) Forward and Left motion. Kind of like what happens in a PWM.

Download RoboAuto here:
Get it on Google Play
Google Play and the Google Play logo are trademarks of Google LLC.

This Post therefore will be updated soon and is worth a bookmark.
UPDATE (22/October/17): Savvy has been renamed to RoboAuto and published on Google Playstore
UPDATE (10/August/16) : Few Screenshots and content have been updated!
UPDATE (25/August/16 ):  Savvy(Now RoboAuto) is available for testing and debugging for FREE
UPDATE (12/September/16): Added Internet Access permission for Google Form Integration.
UPDATE (19/March/17): Fixed a lot of bugs in the layout scale. Renamed "Turn Light" to "Rear Light."

Do report any bugs you encounter.
Credits (19/March/17):  Thank You Shivani for your Feedback on Design Issues

Good Stuff Take Time.

Used:
Android Studio 2.3.3
Lenovo S6600 , Micromax Knight Cameo for Debugging.

Written Sunday, August 07, 2016 by Sanjeev Tripathi

 , , , ,  No comments

Programming and Simulation of AVR with HC-05: PART II

The final build of my work with HC-05 involves using AVR ATmega32 microcontroller which is a powerful micro controller. The Controller is programmed in AVR GCC.

For HC-05 (or any other serial communication system), the USART needs to be set on the ATmega32 through programming.
Here is a snippet of code being used for UART in my build:

Setting up UART


The Primary task of responding to various commands is done by a switch statement which is way easier to modify and work on instead of  IF-ELSE statements.Since Each character has a unique ASCII Code, we have a lot of commands which can be given to the micro controller.For example ,there are 26 lower case + 26 upper case + 10 digits + about 32 symbols which equal about 94 unique commands.
Snippet of code for Switch:

Decision Task by Switch Statement

The character 'ch' is set to receive value from registers whenever the receive interrupt is triggered by the HC-05.The ISR (Interrupt Service Routine) is defined as follows:

Interrupt Service Routine

Simulation of code is done in Proteus 8.4 using all the peripherals included. Input here is given from a virtual terminal however HC-05 would have served the same objective.

Simulation in Proteus

A video accompanying the simulation conveys that the code works fine and can be implemented into the hardware.Here's the vid :


Used:
Proteus 8.4
Atmel Studio 7.0

Written Friday, July 22, 2016 by Sanjeev Tripathi

 , ,  1 comment

A Simple Edge Detection Algorithm

In tasks of image processing, finding edges in an image occasionally serves as the basis for the evaluations.

However, imaging functions usually take up a lot of memory and computation power due to the complexity of  the equations which limit their usage on low memory devices like microcontrollers and single chip boards.

In such case a Linear equation can serve our purpose with a low computation complexity, efficient memory utilisation and faster execution time.

Loading image to find edges :

Self advertisement at its best ;)
Reading and displaying image to process


Edge finding algorithm with intuitive coefficient representation:

Edge Finder 

Edges obtained by the difference equation:

Processed image overlayed on original image

Here is the Edgefinder function, Lines 8-10 are what we are supposed to be looking at :


1:  function out=Edgefinder(I)  
2:    I1=rgb2gray(I);      % conversion to grayscale  
3:    I2=imbinarize(I1);   % conversion to binary image,using Otsu's method  
4:    [m,n]=size(I2);      % size of the image  
5:    out=zeros(size(I2)); % preallocating memory  
6:    for i=2:m-1  
7:      for j=2:n-1  
8:         out(i,j)=  (0.25*I2(i+1,j+1))-   I2(i+1,j) + (0.25*I2(i+1,j-1)) ...  
9:                        - I2(i,j+1)   + 3*I2(i,j)   -       I2(i,j-1) ...  
10:                  +(0.25*I2(i-1,j+1))-   I2(i-1,j) + (0.25*I2(i-1,j-1));  
11:      end  
12:    end  
13:  out=imoverlay(I,out,'red');  %  'I' was preserved for image overlay,not really necessary  
14:  imshow(out);  
15:  end 

Take a look at the coefficients of the difference equation:





Observe the oscillatory nature of the coefficients: A characteristics of a high pass filter. 

Also you may observe that the sum of all coefficients is equal to Zero,

quite intuitive.



It is also evident that this algorithm can find edges at all inclinations so simplicity and efficiency are the key factors here.



Peace out.



Used:

MATLAB 2016a












Written Tuesday, July 19, 2016 by Sanjeev Tripathi

 , , 

 No comments









            

          

          

        

          

        











Starting up with HC-05 Bluetooth Module : PART I









Recent aspiration of mine has been to develop on both AVR and Android a full fledged package for Bluetooth communication







For Prototyping I chose this: 











Setup on this:












With these: 









Looking Something Like this: 









Configured using this:









And Running This:









SO FAR SO GOOD :)














Written Thursday, July 14, 2016 by Sanjeev Tripathi

 , , , 

 No comments









            

          

          

        

          

        











Quick Start Button for Jupyter Notebook









Running Jupyter from the CMD all the time on a windows machine is quite tedious and time consuming.



You gotta navigate to your prefered directory ,then call up CMD (or vice versa) and then execute the command "Jupyter Notebook"



Sure we can write a python program to do that for us, I prefer to use Windows Command Shell for such stuff. Its easy, quick and gets the job done everytime!



Take a look at the script below, only the last two lines are crucial. 




 @echo off  
 title Jupyter_Jumpstart  
 color 70  
 cd C:\Users\Sanjeev\Documents\Jupyter Notebook\  
 jupyter notebook  






Copy Paste the above code and save the notepad file as "anything.bat" (exclude the commas).



You are free to choose anything for the Title, Color and Location of your notebook.








I kept the script on desktop among others softwares. Double click and you are good to go!





Most prominent benefit of this method that I found is that all your .ipynb files are stored at a location  of your choice thus no chance of misplacing your notebooks



Peace.












Written Thursday, June 30, 2016 by Sanjeev Tripathi

 , , 

 No comments









            

          

          

        

          

        











Linear Regression using Gradient Descent









Gradient is the generalization of the usual concept of derivative to a function of several variables. Err... Simply put, the variation of a parameter in  an environment  such that measuring any two points inform us about the direction of increasing density.



Descent is well.. descent ..on the gradient. That is to reach the highest (or lowest?) density in gradual steps.



The task of Gradient Descent algorithm is to find t0 and t1 such that y = t0 +  (t1 * x)

For more, go here Gradient Descent








Input Process








Gradient Descent 



Observe that :

           if x is input , 

              then

                  output   y= 3x

This is learnt by the algorithm through the input training data.   





And the fitting parameter  t1 also gradually converges to ~3








Regression on new input 





The machine learns that the output needs to be 3 times that of input just by learning from data.

Well documented Code is on GitHub



Due to limited buffer space, all iterations with their respective parameters are logged into a data file.Here is the data file for the above set:




Everything done in Visual Studio 2010












Written Saturday, June 25, 2016 by Sanjeev Tripathi

 

 No comments









        

      












Subscribe to:
Posts (Atom)