Auto Mark - Face Tracked Attendance

Made by Snehashish Laskar

---

Auto Mark is a tool that uses facial recognition and computer vision to process images and video input and keep track of student's movements from classroom to classroom.

The system assumes a central server running the Postgres SQL Database and the web interface written flask alongside multiple smaller machines installed at different rooms where the device side scripts will be executed.

Server Side Preperations

Postegres SQL Database Setup

First step is to install and ready Postgres SQL on the server. This can be done as follows:

sudo apt update
sudo apt upgrade

sudo apt install postgresql postgresql-contrib

Next check if the installation is working and running properly.

sudo systemctl status postgresql

Next we login using the default postgress account and create a new operation user.

sudo -i -u postgres
psql

Once we are in the postgres shell, we execute the following commands to crete the new user and new database.

CREATE USER myuser WITH PASSWORD 'mypassword';
CREATE DATABASE auto_mark OWNER myuser;
GRANT ALL PRIVILEGES ON DATABASE auto_mark TO myuser;

Next we run the following sql script to setup the database.

-- SEQUENCES
CREATE SEQUENCE public.admin_auth_id_seq START WITH 1 INCREMENT BY 1;
CREATE SEQUENCE public.face_profiles_id_seq START WITH 1 INCREMENT BY 1;
CREATE SEQUENCE public.sightings_id_seq START WITH 1 INCREMENT BY 1;

-- TABLE: admin_auth
CREATE TABLE public.admin_auth (
    id integer NOT NULL DEFAULT nextval('public.admin_auth_id_seq'),
    user_id text NOT NULL,
    password text NOT NULL,
    auth_level integer NOT NULL,
    PRIMARY KEY (id)
);

-- TABLE: face_profiles
CREATE TABLE public.face_profiles (
    id integer NOT NULL DEFAULT nextval('public face_profiles_id_seq'),
    name text NOT NULL,
    encoding double precision[] NOT NULL,
    PRIMARY KEY (id)
);

-- TABLE: sightings
CREATE TABLE public.sightings (
    id integer NOT NULL DEFAULT nextval('public.sightings_id_seq'),
    name text NOT NULL,
    date date NOT NULL,
    "time" time without time zone NOT NULL,
    room text NOT NULL,
    PRIMARY KEY (id)
);

Setup The Admin Web Server

To setup the environment for the web server we need python 3.10 or greater. We can setup a virtual environment to install all the dependencies.

python3 -m venv .env
source .env/bin/activate
pip3 install -r requirements.txt

Then we create a config.py module to store the database settings

DB_HOST = ""
DB_USERNAME = ""
DB_PASSWORD = ""
DB_NAME = ""

from pathlib import Path
import os

DIRECTORY = str(Path(__file__).resolve().parent)
if not os.path.exists(DIRECTORY+"/data_set"):
    os.mkdir(DIRECTORY+"/data_set")

Then, create an admin user by running the file interact.py

python3 interact.py --add-admin

Finally, to run the server we just need to run the following command and it will start a default flask wsgi server on your local network.

python3 server.py

Client Side Preperations

Assuming we have some sort of operating installed on all the devices while being equiped with a web camera, we can setup a local python environment and run the face recognition program.

To setup the environment for the face recognition client we need python 3.10 or greater. We can setup a virtual environment to install all the dependencies.

python3 -m venv .env
source .env/bin/activate
pip3 install -r requirements.txt

Next, we create a config.py module to store the database settings. The variable ROOM NUMBER will indicate which device the requests are coming from.

DB_HOST = ""
DB_USERNAME = ""
DB_PASSWORD = ""
DB_NAME = ""
ROOM_NUMBER = ""

from pathlib import Path
import os

DIRECTORY = str(Path(__file__).resolve().parent)
if not os.path.exists(DIRECTORY+"/data_set"):
    os.mkdir(DIRECTORY+"/data_set")

Then to start the program we use the following command:

python3 device.py

User Interface