Skip to content

Latest commit

 

History

History
112 lines (85 loc) · 5.7 KB

README.md

File metadata and controls

112 lines (85 loc) · 5.7 KB

EasyOCR-cpp

alt text

Custom C++ implementation of EasyOCR

Built and tested on Windows 11, libtorch 1.13+cpu and OpenCV 4.6

This C++ project implements the pre/post processing to run a OCR pipeline consisting of a text detector CRAFT, and a CRNN based text recognizer. Unlike the EasyOCR python which is API based, this repo provides a set of classes to show how you can integrate OCR in any C++ program for maximum flexibility. The torchExample.cpp main program highlights how to utilize all elements of the EasyOCR-cpp pipeline. Because a test program is only provided, make sure to configure your input image within torchExample.cpp if you only plan to utilize the test program.

thrown together quickly within a week so many debug comments :)

Libtorch is being utilized with an in-house class I usually use for C++ inference TorchModel, and OpenCV for the pre/post processing steps. The TorchModel class can easily adapted to run inference on most Pytorch models converted to TorchScript. Provides some handy functions to pre process opencv::Mat and handle device usage (GPU,CPU). Great starting point for C++ based inference for Pytorch Models.

Some features that have yet to be implemented:

  • beam search, only implemented greedy decoding
  • .txt/.pdf output
  • exact bounding box merge alg from EasyOCR, opted for custom one which is less complex
  • support for other languages, atm only english is supported.

If you would like to support feel free to make a PR, or a issue if you are having trouble.

Setup with Docker (Thanks to @BBO-repo )

Docker container with VSCode

To build and run through visual studio code, make sure the remote development extension is installed.

The repository provides a .devcontainer.json which should allow you to directly reopen the cloned folder in a docker container using the bottom left icon and select Reopen in containeras illustrated below vs-open-in-container

When the container is finally built, you can open a terminal and your working folder should be /workspace/ocr-c++ as mentionned .devcontainer.json the field "workspaceFolder": "/workspaces/ocr-c++". You should have the two folders:

  • thirdparty containing the automatically downloaded libtorch library
  • EasyOCR-cpp containing the code

You can proceed as usual to build the cmake project

mkdir -p /workspaces/ocr-c++/EasyOCR-cpp/build
cd /workspaces/ocr-c++/EasyOCR-cpp/build
cmake ..
make

The binaries should be available in build folder, to run the example application just run:

cd /workspaces/ocr-c++/EasyOCR-cpp/build
./torchTest

Docker container with command line

Considering that the repository was cloned in the folder /my/working/directory/EasyOCR-cpp for illustration.

Build the docker container

Build a docker image named for example ocr_engine from the Dockerfile inside the /my/working/directory/EasyOCR-cpp directory.

docker build --progress=plain -t ocr_engine /my/working/directory/EasyOCR-cpp

Build a container named for example EasyOCR-cpp from the ocr_engine generated image

docker create --name EasyOCR-cpp --mount type=bind,source=/my/working/directory/EasyOCR-cpp,target=/workspaces/ocr-c++/EasyOCR-cpp ocr_engine:latest

You can now start, stop or restart the generated EasyOCR-cpp container.
From a terminal, enter the following command to start the container and getting a terminal inside the container with an interactive mode

docker container start EasyOCR-cpp
docker exec -ti EasyOCR-cpp bash

You should have a bash terminal inside the container, entering pwd and ls commands should output the following:

pwd
  /workspaces/ocr-c++
ls
  thirdparty EasyOCR-cpp

Similarly to previously to build cmake project, go to source code folder cd /workspaces/ocr-c++/EasyOCR-cpp, make a folder build and go inside, then cmake .. and make

cd /workspaces/ocr-c++/EasyOCR-cpp
mkdir build
cd build
cmake ..
make

Then similarly, the binaries should be available in build folder, to run the example application just run:

cd /workspaces/ocr-c++/EasyOCR-cpp/build
./torchTest

Setup to run without Docker

Dependencies

Click to Download libtorch - > download

Use OpenCV Windows installer and unzip (v4.6) - > OpenCV libs

Make sure to change the location in the Makefile for OpenCV to point to your OpenCV build dir

Create a build directory within the repo, cd to it and run

cmake -DCMAKE_PREFIX_PATH=<absolute-path-to-libtorch-folder> ..

This will generate a solution within the build folder you can open up in Visual Studio. Make sure to use the Release config when building

Running

Configure your input image here. Currently the test program is using the test.jpg which comes in the repo.

Launch from command-line, or within Visual Studio after building.

Since its designed to be used in a C++ program, text is not being written to disk at the moment An output image will be generated in the main repo dir containing an annotated version of the input image with detection bounding boxes