Fetch the repository succeeded.
This project is a C + + code modification and encapsulation based on Baidu PaddleOCR Net tool class library. It includes the table recognition function of text recognition, text detection and statistical analysis based on text detection results. At the same time, it is optimized to improve the recognition accuracy in the case of inaccurate small image recognition. It contains ultra lightweight Chinese OCR with a total model of only 8.6M size. The single model supports Chinese and English digit combination recognition, vertical text recognition and long text recognition. Support multiple text detection at the same time.
The project encapsulation is extremely simplified, and the actual call is only a few lines of code, which greatly facilitates the use of middle and downstream developers and reduces the entry level of paddleocr. At the same time, different functions are provided Net framework to facilitate application development and deployment in various industries. Nuget package is a high-precision Chinese and English OCR that can be installed and used immediately, can be deployed offline, and can be recognized without network.
Paddleocr DLL file is a C + + dynamic library modified from the C + + code of the open source project PaddleOCR and compiled based on x64 of OpenCV.
This project can only be compiled and used on x64 CPU, so 32-bit is not supported.
Linux platform is not supported for the time being. If there are cross platform requirements, please refer to systen Drawing. dll、Systen. Drawing. Common. DLL reference is deleted and recompiled.
The project currently supports the following net frameworks:
net35;net40;net45;net451;net452;net46;net461;net462;net47;net471;net472;net48;
netstandard2.0;netcoreapp3.1;
net5.0;net6.0;
OCR recognition model library supports both official models and self-trained models. Bridge completely according to the OCR interface.
The project deploys its own lightweight version of 8.6m model library and server version model library (more accurate and need to be downloaded). The model library can be changed to meet the actual needs.
PaddleOCR models download address
If it needs to be modified into a server version model library, the reference code is as follows: (assuming that the server version model library is under the folder inforenceserver of the running directory)
OpenFileDialog ofd = new OpenFileDialog();
ofd.Filter = "*.*|*.bmp;*.jpg;*.jpeg;*.tiff;*.tiff;*.png";
if (ofd.ShowDialog() != DialogResult.OK) return;
var imagebyte = File.ReadAllBytes(ofd.FileName);
Bitmap bitmap = new Bitmap(new MemoryStream(imagebyte));
//With lightweight Chinese and English model
// OCRModelConfig config = null;
//Server Chinese and English model
//OCRModelConfig config = new OCRModelConfig();
//string root = Environment.CurrentDirectory;
//string modelPathroot = root + @"\inferenceserver";
//config.det_infer = modelPathroot + @"\ch_ppocr_server_v2.0_det_infer";
//config.cls_infer = modelPathroot + @"\ch_ppocr_mobile_v2.0_cls_infer";
//config.rec_infer = modelPathroot + @"\ch_ppocr_server_v2.0_rec_infer";
//config.keys = modelPathroot + @"\ppocr_keys.txt";
//English and digital models
OCRModelConfig config = new OCRModelConfig();
string root = Environment.CurrentDirectory;
string modelPathroot = root + @"\en";
config.det_infer = modelPathroot + @"\ch_PP-OCRv2_det_infer";
config.cls_infer = modelPathroot + @"\ch_ppocr_mobile_v2.0_cls_infer";
config.rec_infer = modelPathroot + @"\en_number_mobile_v2.0_rec_infer";
config.keys = modelPathroot + @"\en_dict.txt";
OCRParameter oCRParameter = new OCRParameter ();
OCRResult ocrResult = new OCRResult();
//It is suggested that the program can be initialized once globally. It is not necessary to initialize every time of identification, which is easy to report errors.
PaddleOCREngine engine = new PaddleOCREngine(config, oCRParameter);
{
ocrResult = engine.DetectText(bitmap );
}
if (ocrResult != null)
{
MessageBox.Show(ocrResult.Text,"Recognition results");
}
//When OCR is no longer used, please release paddlecrengine
C++Download address of prediction Library for windows
[PaddleOCR official parameters]](https://gitee.com/paddlepaddle/PaddleOCR/tree/release/2.4/deploy/cpp_infer)
PaddleOCR //PaddleOCR.dll source code folder
|--cpp //PaddleOCR.dll Cpp source code
|--include //PaddleOCR.dll .h source code
PaddleOCRLib // the files to need on the runtime
|--inference // lite models for the Chinese characters
|--libiomp5md.dll //
|--mkldnn.dll //
|--mklml.dll //
|--opencv_world411.dll //
|--paddle_inference.dll //
|--PaddleOCR.dll //
PaddleOCRSharp //.NET library
PaddleOCRDemo //Demo folder
|--Cpp //Cpp folder
|--PaddleOCRCppDemo //C++ demo
|--PaddleOCRSharpDemo //.NET demo
|--WebAPIDemo //.NET WebAPI demo
As for the source code compilation, it is recommended to use vs2022 version. If you cannot compile, please switch to release and then switch back to debug。
If you cannot compile due to the problem of framework compilation, please modify paddleocrsharp \ paddleocrsharp Csproj file, delete the frame not available on the current computer, See Microsoft documentation for specific framework descriptionTarget framework in SDK style project
<TargetFrameworks>
net35;net40;net45;net451;net452;net46;net461;net462;net47;net471;net472;net48;
netstandard2.0;netcoreapp3.1;
net5.0;net6.0;
</TargetFrameworks>
OpenFileDialog ofd = new OpenFileDialog();
ofd.Filter = "*.*|*.bmp;*.jpg;*.jpeg;*.tiff;*.tiff;*.png";
if (ofd.ShowDialog() != DialogResult.OK) return;
var imagebyte = File.ReadAllBytes(ofd.FileName);
Bitmap bitmap = new Bitmap(new MemoryStream(imagebyte));
OCRModelConfig config = null;
OCRParameter oCRParameter = new OCRParameter ();
OCRResult ocrResult = new OCRResult();
//It is suggested that the program can be initialized once globally. It is not necessary to initialize every time of identification, which is easy to report errors。
PaddleOCREngine engine = new PaddleOCREngine(config, oCRParameter);
{
ocrResult = engine.DetectText(bitmap );
}
if (ocrResult != null)
{
MessageBox.Show(ocrResult.Text,"results");
}
此处可能存在不合适展示的内容，页面不予展示。您可通过相关编辑功能自查并修改。
如您确认内容无涉及 不当用语 / 纯广告导流 / 暴力 / 低俗色情 / 侵权 / 盗版 / 虚假 / 无价值内容或违法国家有关法律法规的内容，可点击提交进行申诉，我们将尽快为您处理。
Activity
Community
Health
Trend
Influence
：Code submit frequency
：React/respond to issue & PR etc.
：Well-balanced team members and collaboration
：Recent popularity of project
：Star counts, download counts etc.