C# OnnxRuntime 实现百度网盘AI大赛-表格检测第2名方案第二部分-表格分割

目录

说明

效果

模型

项目

代码

参考

下载

其他

说明

百度网盘AI大赛-表格检测的第2名方案。

该算法包含表格边界框检测、表格分割和表格方向识别三个部分，首先，ppyoloe-plus-x 对边界框进行预测，并对置信度较高的表格边界框（box）进行裁剪。裁剪后的单个表格实例会送入到DBNet中进行语义分割，分割结果通过opencv轮廓处理获得表格关键点(point)。之后，我们根据DBNet计算的关键点在裁剪后的单个表格实例上绘制表格边界。最后，PP-LCNet结合表格边界先验和表格实例图像，对表格的方向进行预测，并根据之前定义的几何轮廓点与语义轮廓点的对应关系，将几何轮廓点映射为语义轮廓点。

本文使用C# OnnxRuntime 实现百度网盘AI大赛-表格检测第2名方案第二部分-表格分割

效果

模型

Model Properties
-------------------------
date：2024-10-28T08:16:43.725877
description：Ultralytics YOLO11l-seg model trained on coco-seg.yaml
author：Ultralytics
version：8.3.23
task：segment
license：AGPL-3.0 License (https://ultralytics.com/license)
docs：https://docs.ultralytics.com
stride：32
batch：1
imgsz：[800, 800]
names：{0: 'table'}
---------------------------------------------------------------

Inputs
-------------------------
name：images
tensor：Float[1, 3, 800, 800]
---------------------------------------------------------------

Outputs
-------------------------
name：output0
tensor：Float[1, 37, 13125]
name：output1
tensor：Float[1, 32, 200, 200]
---------------------------------------------------------------

项目

代码

using Microsoft.ML.OnnxRuntime;
using Microsoft.ML.OnnxRuntime.Tensors;
using OpenCvSharp;
using System;
using System.Collections.Generic;
using System.ComponentModel;
using System.Data;
using System.Drawing;
using System.Linq;
using System.Security.Cryptography;
using System.Text;
using System.Web;
using System.Windows.Forms;
using static System.Net.Mime.MediaTypeNames;

namespace Onnx_Yolov8_Demo
{
    public partial class Form1 : Form
    {
        public Form1()
        {
            InitializeComponent();
        }

        string fileFilter = "*.*|*.bmp;*.jpg;*.jpeg;*.tiff;*.tiff;*.png";
        string image_path = "";
        string startupPath;
        string classer_path;
        DateTime dt1 = DateTime.Now;
        DateTime dt2 = DateTime.Now;
        string model_path;
        Mat image;
        SegmentationResult result_pro;
        Mat result_image;

        SessionOptions options;
        InferenceSession onnx_session;
        Tensor input_tensor;
        List input_container;
        IDisposableReadOnlyCollection result_infer;
        DisposableNamedOnnxValue[] results_onnxvalue;

        Tensor result_tensors_det;
        Tensor result_tensors_proto;

        private void button1_Click(object sender, EventArgs e)
        {
            OpenFileDialog ofd = new OpenFileDialog();
            ofd.Filter = fileFilter;
            if (ofd.ShowDialog() != DialogResult.OK) return;
            pictureBox1.Image = null;
            image_path = ofd.FileName;
            pictureBox1.Image = new Bitmap(image_path);
            textBox1.Text = "";
            image = new Mat(image_path);
            pictureBox2.Image = null;
        }

        private void button2_Click(object sender, EventArgs e)
        {
            if (image_path == "")
            {
                return;
            }

            button2.Enabled = false;

           System.Windows.Forms.Application.DoEvents();

            // 配置图片数据
            image = new Mat(image_path);
            int max_image_length = image.Cols > image.Rows ? image.Cols : image.Rows;
            Mat max_image = Mat.Zeros(new OpenCvSharp.Size(max_image_length, max_image_length), MatType.CV_8UC3);
            Rect roi = new Rect(0, 0, image.Cols, image.Rows);
            image.CopyTo(new Mat(max_image, roi));

            float[] det_result_array = new float[37 * 13125];
            float[] proto_result_array = new float[32 * 200 * 200];
            float[] factors = new float[4];
            factors[0] = factors[1] = (float)(max_image_length / 800.0);
            factors[2] = image.Rows;
            factors[3] = image.Cols;

            // 将图片转为RGB通道
            Mat image_rgb = new Mat();
            Cv2.CvtColor(max_image, image_rgb, ColorConversionCodes.BGR2RGB);
            Mat resize_image = new Mat();
            Cv2.Resize(image_rgb, resize_image, new OpenCvSharp.Size(800, 800));

            // 输入Tensor
            for (int y = 0; y < resize_image.Height; y++)
            {
                for (int x = 0; x < resize_image.Width; x++)
                {
                    input_tensor[0, 0, y, x] = resize_image.At(y, x)[0] / 255f;
                    input_tensor[0, 1, y, x] = resize_image.At(y, x)[1] / 255f;
                    input_tensor[0, 2, y, x] = resize_image.At(y, x)[2] / 255f;
                }
            }

            //将 input_tensor 放入一个输入参数的容器，并指定名称
            input_container.Add(NamedOnnxValue.CreateFromTensor("images", input_tensor));

            dt1 = DateTime.Now;
            //运行 Inference 并获取结果
            result_infer = onnx_session.Run(input_container);

            dt2 = DateTime.Now;

            // 将输出结果转为DisposableNamedOnnxValue数组
            results_onnxvalue = result_infer.ToArray();

            // 读取第一个节点输出并转为Tensor数据
            result_tensors_det = results_onnxvalue[0].AsTensor();
            result_tensors_proto = results_onnxvalue[1].AsTensor();

            det_result_array = result_tensors_det.ToArray();
            proto_result_array = result_tensors_proto.ToArray();

            resize_image.Dispose();
            image_rgb.Dispose();

            result_pro = new SegmentationResult(classer_path, factors);
            result_image = result_pro.draw_result(result_pro.process_result(det_result_array, proto_result_array), image.Clone());

            if (!result_image.Empty())
            {
                pictureBox2.Image = new Bitmap(result_image.ToMemoryStream());
                textBox1.Text = "推理耗时:" + (dt2 - dt1).TotalMilliseconds + "ms";
            }
            else
            {
                textBox1.Text = "无信息";
            }

            button2.Enabled = true;
        }

        private void Form1_Load(object sender, EventArgs e)
        {
            startupPath = System.Windows.Forms.Application.StartupPath;
            model_path = startupPath + "\\yolo_edge_det.onnx";
            classer_path = startupPath + "\\lable.txt";

            // 创建输出会话，用于输出模型读取信息
            options = new SessionOptions();
            options.LogSeverityLevel = OrtLoggingLevel.ORT_LOGGING_LEVEL_INFO;
            // 设置为CPU上运行
            options.AppendExecutionProvider_CPU(0);

            // 创建推理模型类，读取本地模型文件
            onnx_session = new InferenceSession(model_path, options);//model_path 为onnx模型文件的路径

            // 输入Tensor
            input_tensor = new DenseTensor(new[] { 1, 3, 800, 800 });

            // 创建输入容器
            input_container = new List();

            image_path = "1.jpg";
            pictureBox1.Image = new Bitmap(image_path);

        }
    }
}
 

  1. using Microsoft.ML.OnnxRuntime;
  2. using Microsoft.ML.OnnxRuntime.Tensors;
  3. using OpenCvSharp;
  4. using System;
  5. using System.Collections.Generic;
  6. using System.ComponentModel;
  7. using System.Data;
  8. using System.Drawing;
  9. using System.Linq;
  10. using System.Security.Cryptography;
  11. using System.Text;
  12. using System.Web;
  13. using System.Windows.Forms;
  14. using static System.Net.Mime.MediaTypeNames;
  15.  
  16. namespace Onnx_Yolov8_Demo
  17. {
  18.     public partial class Form1 : Form
  19.     {
  20.         public Form1()
  21.         {
  22.             InitializeComponent();
  23.         }
  24.  
  25.         string fileFilter = "*.*|*.bmp;*.jpg;*.jpeg;*.tiff;*.tiff;*.png";
  26.         string image_path = "";
  27.         string startupPath;
  28.         string classer_path;
  29.         DateTime dt1 = DateTime.Now;
  30.         DateTime dt2 = DateTime.Now;
  31.         string model_path;
  32.         Mat image;
  33.         SegmentationResult result_pro;
  34.         Mat result_image;
  35.  
  36.         SessionOptions options;
  37.         InferenceSession onnx_session;
  38.         Tensor<float> input_tensor;
  39.         List<NamedOnnxValue> input_container;
  40.         IDisposableReadOnlyCollection<DisposableNamedOnnxValue> result_infer;
  41.         DisposableNamedOnnxValue[] results_onnxvalue;
  42.  
  43.         Tensor<float> result_tensors_det;
  44.         Tensor<float> result_tensors_proto;
  45.  
  46.         private void button1_Click(object sender, EventArgs e)
  47.         {
  48.             OpenFileDialog ofd = new OpenFileDialog();
  49.             ofd.Filter = fileFilter;
  50.             if (ofd.ShowDialog() != DialogResult.OK) return;
  51.             pictureBox1.Image = null;
  52.             image_path = ofd.FileName;
  53.             pictureBox1.Image = new Bitmap(image_path);
  54.             textBox1.Text = "";
  55.             image = new Mat(image_path);
  56.             pictureBox2.Image = null;
  57.         }
  58.  
  59.         private void button2_Click(object sender, EventArgs e)
  60.         {
  61.             if (image_path == "")
  62.             {
  63.                 return;
  64.             }
  65.  
  66.             button2.Enabled = false;
  67.  
  68.            System.Windows.Forms.Application.DoEvents();
  69.  
  70.             // 配置图片数据
  71.             image = new Mat(image_path);
  72.             int max_image_length = image.Cols > image.Rows ? image.Cols : image.Rows;
  73.             Mat max_image = Mat.Zeros(new OpenCvSharp.Size(max_image_length, max_image_length), MatType.CV_8UC3);
  74.             Rect roi = new Rect(0, 0, image.Cols, image.Rows);
  75.             image.CopyTo(new Mat(max_image, roi));
  76.  
  77.             float[] det_result_array = new float[37 * 13125];
  78.             float[] proto_result_array = new float[32 * 200 * 200];
  79.             float[] factors = new float[4];
  80.             factors[0] = factors[1] = (float)(max_image_length / 800.0);
  81.             factors[2] = image.Rows;
  82.             factors[3] = image.Cols;
  83.  
  84.             // 将图片转为RGB通道
  85.             Mat image_rgb = new Mat();
  86.             Cv2.CvtColor(max_image, image_rgb, ColorConversionCodes.BGR2RGB);
  87.             Mat resize_image = new Mat();
  88.             Cv2.Resize(image_rgb, resize_image, new OpenCvSharp.Size(800, 800));
  89.  
  90.             // 输入Tensor
  91.             for (int y = 0; y < resize_image.Height; y++)
  92.             {
  93.                 for (int x = 0; x < resize_image.Width; x++)
  94.                 {
  95.                     input_tensor[0, 0, y, x] = resize_image.At<Vec3b>(y, x)[0] / 255f;
  96.                     input_tensor[0, 1, y, x] = resize_image.At<Vec3b>(y, x)[1] / 255f;
  97.                     input_tensor[0, 2, y, x] = resize_image.At<Vec3b>(y, x)[2] / 255f;
  98.                 }
  99.             }
  100.  
  101.             //将 input_tensor 放入一个输入参数的容器，并指定名称
  102.             input_container.Add(NamedOnnxValue.CreateFromTensor("images", input_tensor));
  103.  
  104.             dt1 = DateTime.Now;
  105.             //运行 Inference 并获取结果
  106.             result_infer = onnx_session.Run(input_container);
  107.  
  108.             dt2 = DateTime.Now;
  109.  
  110.             // 将输出结果转为DisposableNamedOnnxValue数组
  111.             results_onnxvalue = result_infer.ToArray();
  112.  
  113.             // 读取第一个节点输出并转为Tensor数据
  114.             result_tensors_det = results_onnxvalue[0].AsTensor<float>();
  115.             result_tensors_proto = results_onnxvalue[1].AsTensor<float>();
  116.  
  117.             det_result_array = result_tensors_det.ToArray();
  118.             proto_result_array = result_tensors_proto.ToArray();
  119.  
  120.             resize_image.Dispose();
  121.             image_rgb.Dispose();
  122.  
  123.             result_pro = new SegmentationResult(classer_path, factors);
  124.             result_image = result_pro.draw_result(result_pro.process_result(det_result_array, proto_result_array), image.Clone());
  125.  
  126.             if (!result_image.Empty())
  127.             {
  128.                 pictureBox2.Image = new Bitmap(result_image.ToMemoryStream());
  129.                 textBox1.Text = "推理耗时:" + (dt2 - dt1).TotalMilliseconds + "ms";
  130.             }
  131.             else
  132.             {
  133.                 textBox1.Text = "无信息";
  134.             }
  135.  
  136.             button2.Enabled = true;
  137.         }
  138.  
  139.         private void Form1_Load(object sender, EventArgs e)
  140.         {
  141.             startupPath = System.Windows.Forms.Application.StartupPath;
  142.             model_path = startupPath + "\\yolo_edge_det.onnx";
  143.             classer_path = startupPath + "\\lable.txt";
  144.  
  145.             // 创建输出会话，用于输出模型读取信息
  146.             options = new SessionOptions();
  147.             options.LogSeverityLevel = OrtLoggingLevel.ORT_LOGGING_LEVEL_INFO;
  148.             // 设置为CPU上运行
  149.             options.AppendExecutionProvider_CPU(0);
  150.  
  151.             // 创建推理模型类，读取本地模型文件
  152.             onnx_session = new InferenceSession(model_path, options);//model_path 为onnx模型文件的路径
  153.  
  154.             // 输入Tensor
  155.             input_tensor = new DenseTensor<float>(new[] { 1, 3, 800, 800 });
  156.  
  157.             // 创建输入容器
  158.             input_container = new List<NamedOnnxValue>();
  159.  
  160.             image_path = "1.jpg";
  161.             pictureBox1.Image = new Bitmap(image_path);
  162.  
  163.         }
  164.     }
  165. }

参考

https://github.com/hpc203/TableDetection

下载

源码下载

其他

C# OpenCvSharp DNN 第一部分-表格边界框检测-CSDN博客

C# OpenCvSharp DNN 第三部分-表格方向识别-CSDN博客

天天代码码天天
微信公众号
.NET 人工智能实践
注：本文转载自blog.csdn.net的天天代码码天天的文章"https://lw112190.blog.csdn.net/article/details/144487874"。版权归原作者所有，此博客不拥有其著作权，亦不承担相应法律责任。如有侵权，请联系我们删除。
复制链接
相关推荐
发表评论

评论记录：

未查询到任何数据！