# PaddleOCRSharp-VL-1.5

**Repository Path**: idkook/PaddleOCRSharp-VL-1.5

## Basic Information

- **Project Name**: PaddleOCRSharp-VL-1.5
- **Description**: PaddleOCRSharp-VL-1.5 is based on the PaddleOCR-VL-1.5 model and utilizes the llama framework's encapsulation library, supports:OCR;Table Recognition;Formula Recognition;Chart Recognition;Spotting;Sea
- **Primary Language**: C#
- **License**: Apache-2.0
- **Default Branch**: master
- **Homepage**: https://gitee.com/raoyutian/PaddleOCRSharp-VL-1.5
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 1
- **Created**: 2026-06-18
- **Last Updated**: 2026-06-18

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# PaddleOCRSharp-VL-1.5

PaddleOCRSharp-VL-1.5 is based on the PaddleOCR-VL-1.5 model and utilizes the llama framework's encapsulation library, supports:OCR;Table Recognition;Formula Recognition;Chart Recognition;Spotting;Seal Recognition

support for LLM for example: DeepSeek、Qwen ...

# How to use

1. Install the Nuget package PaddleOCRSharp-VL-1.5 or download  https://www.nuget.org/packages/PaddleOCRSharp-VL-1.5
 
2. Download the model file,  from the following url:

https://www.modelscope.cn/models/megemini/PaddleOCR-VL-1.5-GGUF

Or LLM model

https://www.modelscope.cn/models/duanzy1/Qwen3.5-9B_guff/files


3. Code

```
using PaddleOCRSharp.VL;

string root = AppContext.BaseDirectory;
root += @"\models\";
//下载模型，假设放在models目录下
string model_file = Path.Combine(root, "PaddleOCR-VL-1.5-GGUF.gguf");
//对话大模型，该字段为空
string mmproj_file = Path.Combine(root, "PaddleOCR-VL-1.5-GGUF-mmproj.gguf");

//创建引擎对象
engine = new PaddleOCRVLEngine();
try
    {
      //加载模型，参数为全路径,mmproj_file对话大模型，该字段为空即可
      engine.LoadModel(model_file, mmproj_file);
      DateTime dt1 = DateTime.Now;
     //加载图片,对话大模型，不用加载图片
     engine.LoadImage("test.jpg")
      //根据提示词，处理图片
      string ocrResult = engine.Generate(Prompt.OCR);
      DateTime dt2 = DateTime.Now;
      var times = (dt2 - dt1).TotalMilliseconds;
      Console.WriteLine($"result:\n{ocrResult}\ntimes:{times}ms");
    }
catch (Exception ex)
    {
      Console.WriteLine($"error:\n{ex.Message}");        
    }
finally
    {
      //按需释放资源
      //engine?.Release();
    }
     Console.ReadKey();

```

4. note

Each Generate does not retain context and historical records



6. If you find it helpful, please feel free to give us a Star and cite our work.

7. 如需定制开发请联系QQ：277784829