License Plate Recognition with go-darknet

Table of Contents

• About
• Requirements
• Installation
  • Get source code
  • Protobuf generation
  • Neural net weights
  • Custom handler
• Usage
  • Server
  • Client

About

This is a gRPC server which accepts image and can make license plate recognition (using YOLOv3 or YOLOv4 neural network).

Server tries to find license plate at first. Then it does OCR (if it's possible).

Neural networks were trained on dataset of russian license plates. But you can train it on another dataset - read about process here AlexeyAB/darknet

Darknet architecture for finding license plates - Yolo V3

Darknet architecture for doing OCR stuff - Yolo V4

No OpenCV installation is needed!

gRPC server accepts this data struct accordion to ODaM specification:

message CamInfo{
    string cam_id = 1; // id of camera (just to identify client app)
    int64 timestamp = 2; // timestamp of vehicle fixation (on client app)
    bytes image = 3; // bytes of full image in PNG-format
    Detection detection = 4; // BBox of detected vehicle (region of interest where License Plate Recognition is needed)
    VirtualLineInfo virtual_line = 5; // Line which detected object has been crossed (not necessary field, but helpfull for real-time detection on road traffic)
}
message Detection{
    int32 x_left = 1;
    int32 y_top = 2;
    int32 height = 3;
    int32 width = 4;
}
message VirtualLineInfo{
    int32 id = 1;
    int32 left_x = 2;
    int32 left_y = 3;
    int32 right_x = 4;
    int32 right_y = 5;
}

Requirements

Please follow instructions from go-darknet. There you will know how to install AlexeyAB's darknet and Go-binding for it.

Instalation

Get source code

Notice: we are using Go-modules

go get https://github.com/LdDl/license_plate_recognition

Download weights and configuration

Notice: please read source code of *.sh script before downloading. This script MAY NOT fit yours needs.

cd cmd/
chmod +x download_data_RU.sh
./download_data_RU.sh

Custom Handler

Do not forget (if needed) to implement AfterFunc

Example is below:

....
rs := &RecognitionServer{
    ....
    AfterFunction: doSomeStuff,
}
....
func doSomeStuff(data *PlateInfo, fileContents []byte) error {
	/*
		If you want, you can implement this function by yourself (and you can wrap this function also)
		Default behaviour: do nothing.
	*/
	return nil
}
....

Usage

Start server

• Navigate to folder with server application source code

  cd cmd/server
  

• Build source code of server application to executable

  go build -o recognition_server main.go
  

• Run server application

  ./recognition_server --cfg conf.toml
  

  Note: Please see conf.toml description for correct usage

Test Client-Server

Notice: server should be started

• Navigate to folder with server application source code

  cd cmd/client
  

• Build source code of client application to executable

  go build -o client_app main.go
  

• Run client application

  ./client_app --host=localhost --port=50051 --file=sample.jpg -x 0 -y 0 --width=4032 --height=3024
  

• Check, if server can handle error (like negative height parameter):

  ./client_app --host=localhost --port=50051 --file=sample.jpg -x 0 -y 0 --width=42 --height=-24
  

• On server's side there will be output something like this:

  2020/06/25 15:31:57
  License plate #0:
      Text: M288HO199
      Deviation (for detected symbols): 1.808632
      Rectangle's borders: (295,1057)-(608,1204)
  License plate #1:
      Text: A100CX777
      Deviation (for detected symbols): 2.295539
      Rectangle's borders: (2049,1384)-(2582,1618)
  Elapsed to find plate and read symbols: 372.108605ms
  

• On server's side the directory './detected' will appear also. Detected license plates will be stored there.