gRPC-based Template Matching and OCR Server with OpenCV

This repository contains a gRPC-based server and client for template matching and Optical Character Recognition (OCR) using OpenCV and Tesseract OCR.

Features

• gRPC server and client for efficient and scalable communication.
• Template matching functionality using OpenCV.
• OCR functionality using Tesseract OCR.
• Cross-platform compatibility.

Dependencies

• gRPC (https://grpc.io/)
• OpenCV (https://opencv.org/)
• Tesseract OCR (https://tesseract-ocr.github.io/)
• Leptonica (http://leptonica.org/)

Demo

Run the server:

./recognizer_async_server
Tesseract Base API Load Successfully
Server listening on 0.0.0.0:50051

Run the client with an image file as an argument:

./recognizer_client img/hey1219.png
OCR RESULT:HEY
1219

Target Image

Server Implementation (recognizer_async_server.cc)

The server is implemented in C++ and exposes two gRPC services: MatchTemplate for template matching and RecognizeOpticalCharacters for OCR. The server initializes Tesseract OCR with the given language data and processes incoming requests.

Client Implementation (recognizer_client.cc)

The client is also implemented in C++ and communicates with the server using gRPC. The client reads the image file, sends it to the server, and receives the OCR result.

Build

1. Install gRPC

2. Install Tesseract

2-1. Tesseract needs below

• Leptonica
• libjpeg
• libtiff
• libpng
• libssl-dev

3. Protoc : generate source files

protoc --proto_path=./protos recognizer.proto --cpp_out=./ --grpc_out=./ --plugin=protoc-gen-grpc=/usr/local/bin/grpc_cpp_plugin

4. Download Tesseract TrainedData

wget https://github.com/tesseract-ocr/tessdata/blob/main/eng.traineddata?raw=true

Put the file tessdata/ or set TESSDATA_PREFIX environment variable.

License

This project is licensed under the terms of the MIT license.