Cargate Project Implementation

The system is built using yolov5 (objects detector) [Reference 1] and PaddlePaddle OCR (license plate recognition) [Reference 2]

It will be used to detect vehicles including Cars and Vans and send the NumberPlate detections over to a FastAPI backend server which will perform the License Plate Recognition (LPR) and allow the access to authorized vehicles.

Run the Backend Server

Before running the project, make sure to first setup the FastAPI backend server, to do so please refer to the following server instructions.

This project can be run on both a PC 💻 as well as on a Jetson Nano 📼

• To Run the Cargate project on a PC, please refer to the following PC instructions.
• To Run the Cargate project on a Jetson Nano, choose from the following two options:
  • Run Cargate on Jetson Nano without Docker
  • Run Cargate on Jetson Nano with Docker

Run Cargate on Jetson Nano without Docker 🐳❌

The Following scripts were tested on a Jetson Nano with JetPack 4.5 image flashed. To download Jetpack 4.5, refer to the following link.

Install torch and torchvision libraries

./install_torch.sh

Install additional libraries

pip3 install -r Carplate-yolov5/requirements-nano.txt

Download weights

gdown https://drive.google.com/uc?id=18tyNWkGC_x9FddZ9hJ5di3_Sc9WPkpd_ -O Carplate-yolov5/yolov5/weights/detection.pt

Run Inference

Detect vehicles using the CSI Camera and send NumberPlate detections over to a FastAPI backend server to perform LPR.

python3 Carplate-yolov5/main.py --show --nano

Run Cargate on Jetson Nano with Docker 🐳

The Following scripts were tested on a Jetson Nano with JetPack 4.5 image flashed. To download Jetpack 4.5, refer to the following link.

Change Docker Default Runtime

This sets "default-runtime": "nvidia" in /etc/docker/daemon.json

./setup.sh

Build the Docker Image

Note: This should take approximately 3-4 hours to complete

./build.sh

Run Inference

Detect vehicles using the CSI Camera and send NumberPlate detections over to a FastAPI backend server to perform LPR.

./run.sh

Demo

External Hardware Circuit Diagram

There are 2 external sensors used in this project:

1. HC-SR04 Ultrasonic Distance Sensor: Used to determine if a vehicle is present before starting the detection. (link)
2. 1 CH Active H/L 5V Relay Module: Used to control the gate and grant access to authorized vehicles. (link)

Appendix

There are 2 ways that you can use to run shell scripts on Linux:

1) By using the bash command:   
$ bash ./scriptname.sh   

2) By making the script executable (To be performed only once):        
$ chmod +x scriptname.sh   

Then run the script normally using the command:   
$ ./scriptname.sh      

References

1. https://github.com/ultralytics/yolov5

2. https://github.com/PaddlePaddle/PaddleOCR

3. https://github.com/otamajakusi/dockerfile-yolov5-jetson