Camera Reporter

This projects handels intelligent monitoring. It uses Darknet AI to classify objects from a USB camera picture, or embedded web cam inside a laptop. If a object is identified, then this web application and send a mail e.g alarm mail with a message to a specific user. Very usefull if you want to have a specific alarm and not just a regular alarm.

Use this software if you want handel intelligent monitoring for security.

Features

• Mail service
• Object identification
• Mobile and tablet suitable
• Login screen with password requirement
• Database storage for emails and messages

Pictures

My desktop with a poor old Dell Precision M6400 from 2007. Yes, it works but it's about 2 seconds delay per image for Yolov4-tiny model.

Mail configuration and message

Darknet files upload

How to install - Ubuntu user

1. Install Java 11, Maven, NodeJS

Java 11

sudo apt-get install openjdk-11-jdk

Maven

sudo apt-get install maven

NodeJS - This is used if you want to work on this project. If you only want to run this project, you don't need NodeJS.

curl -sL https://deb.nodesource.com/setup_14.x | sudo -E bash -
sudo apt-get install -y nodejs

2. Begin first to install MySQL Community Server

sudo apt-get install mysql-server

3. Then create a user e.g myUser with the password e.g myPassword

Login and enter your sudo password or mysql root password

sudo mysql -u root -p

Create user with the host % <-- That's important if you want to access your server from other computers.

CREATE USER 'myUser'@'%' IDENTIFIED BY 'myPassword';

Set the privileges to that user

GRANT ALL PRIVILEGES ON *.* TO 'myUser'@'%';

4. Change your MySQL server so you listening to your LAN address

Open this file

/etc/mysql/mysql.conf.d/mysqld.conf

And change this

bind-address            = 127.0.0.1

To your LAN address where the server is installed on e.g

bind-address            = 192.168.1.34

Then restart your MySQL server

sudo /etc/init.d/mysql restart

If you don't know your LAN address, you can type in this command in linux ifconfig in the terminal

5. Create a Gmail account

Create a Gmail account and go to https://myaccount.google.com/security and enable so you can login from less secure apps. Because Camera-Reporter uses Java Mail to logg into Gmail. This feature exist because if Camera-Reporter is on the fly over a night and something happens, then it will stop everything and send a message back to you.

6. Create Ramdisk for Darknet folder

We need to save our prediction pictures and camera pictures here on RAM memory so we won't harm the harddrive. Create a 200 megabyte Ramdisk for two pictures if you using Yolo V4 Tiny. When applied Yolo V4 Tiny, then you will have about 80 megabytes left. Every time you stard the prediction, then this Camera-Reporter will copy Darknet folder to /mnt/ramdisk

Do the following:

sudo mkdir /mnt/ramdisk
sudo mount -t tmpfs -o rw,size=200M tmpfs /mnt/ramdisk
df -h

Add a Ramdisk automatically at startup Write

sudo nano /etc/fstab

Then paste this line at the bottom

tmpfs  /mnt/ramdisk  tmpfs  rw,size=200M  0   0

Now press Ctrl> + X and then press y and then press Enter to save the file.

If you want to unmount ramdisk folder

sudo umount /mnt/ramdisk

7. Download Camera-Reporter

Download the Camera-Reporter and change the application.properties in the /src/main/resources folder. Here you can set the configuration for your database LAN address, user and password. You can also set a gmail address and its password.

# Database
spring.jpa.show-sql=true
spring.jpa.hibernate.ddl-auto=update
spring.jpa.properties.hibernate.dialect=org.hibernate.dialect.MySQL5Dialect
spring.datasource.url=jdbc:mysql://yourServerIP:3306/CameraReporter?createDatabaseIfNotExist=true&serverTimezone=CET
spring.datasource.username=myUser
spring.datasource.password=myPassword

#Upload
spring.servlet.multipart.max-file-size=500MB
spring.servlet.multipart.max-request-size=500MB

# Mail - Transmitter
mail.host=smtp.gmail.com
mail.port=587
mail.username=yourGMailAddress@gmail.com
mail.password=yourGMailPassword
mail.properties.mail.smtp.auth=true
mail.properties.mail.smtp.starttls.enable=true

# Mail - Reciever
mail.subject = Camera Detection

# Login
spring.security.user.name=myUser
spring.security.user.password=myPassword

8. Run the project

Stand inside of the folder Camera-Reporter and write inside your terminal

mvn spring-boot:run -Pproduction

Now you can go to your web browser and type in the local IP address of the computer there you started this application.

9. Upload Darknet files

Go to https://github.com/AlexeyAB/darknet and download the sourcecode and follow the instructions how to compile under Linux. Then upload the darknet, .data, .cfg, .weights, .names files etc. to the Darknet folder inside this project. There is a YOLO 4 Tiny already included so you can if you want just try this first and see if you get predictions before you doing it any more. Notice that this darknet file included in this project is compiled under Lubuntu Linux 18.04 on a Dell Precision M6400 computer. It has no GPU, CUDA, OPENCV. Just default settings. You don't need OpenCV because this Darknet won't show the predictions.