目录

介绍

效果 

模型信息

项目

代码

下载

---

LaMa Image Inpainting 图像修复 OnnxRuntime-GPU版 Demo

介绍

gihub地址：GitHub - advimman/lama: 🦙 LaMa Image Inpainting, Resolution-robust Large Mask Inpainting with Fourier Convolutions, WACV 2022

🦙 LaMa Image Inpainting, Resolution-robust Large Mask Inpainting with Fourier Convolutions, WACV 2022

效果 

CPU推理效果

GPU推理效果

模型信息

Model Properties
-------------------------
---------------------------------------------------------------

Inputs
-------------------------
name：image
tensor：Float[1, 3, 1000, 1504]
name：mask
tensor：Float[1, 1, 1000, 1504]
---------------------------------------------------------------

Outputs
-------------------------
name：inpainted
tensor：Float[1, 1000, 1504, 3]
---------------------------------------------------------------

项目

安装包及版本如下：

环境：

NVIDIA GeForce RTX 4060 Laptop GPU

cuda12.1+cudnn 8.8.1

代码

using OpenCvSharp;
using System;
using System.Diagnostics;
using System.Drawing;
using System.Windows.Forms;

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

        string fileFilter = "*.*|*.bmp;*.jpg;*.jpeg;*.tiff;*.tiff;*.png";
        string image_path = "";
        string image_path_mask = "";
        string model_path;
        Mat image;
        Mat image_mask;
        LaMa laMa;

        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;
            pictureBox2.Image = null;
            textBox1.Text = "";
            Application.DoEvents();

            image = new Mat(image_path);
            image_mask = new Mat(image_path_mask);

            Stopwatch stopwatch = new Stopwatch();
            stopwatch.Start();

            Mat result = laMa.Run(image,image_mask);

            double costTime = stopwatch.Elapsed.TotalMilliseconds;

            if (chkUseGPU.Checked)
            {
                textBox1.Text = "GPU推理耗时:" + costTime + "ms";
            }
            else {
                textBox1.Text = "CPU推理耗时:" + costTime + "ms";
            }

            if (pictureBox2.Image!=null)
            {
                pictureBox2.Image.Dispose();
            }
            pictureBox2.Image = new Bitmap(result.ToMemoryStream());
            button2.Enabled = true;

            image_mask.Dispose();
            image.Dispose();
        }

        private void Form1_Load(object sender, EventArgs e)
        {
            model_path = "model/big_lama_regular_inpaint.onnx";
            laMa = new LaMa(model_path);

            image_path = "test_img/test.jpg";
            pictureBox1.Image = new Bitmap(image_path);

            image_path_mask = "test_img/mask.jpg";
            pictureBox3.Image = new Bitmap(image_path_mask);
        }

        private void chkUseGPU_CheckedChanged(object sender, EventArgs e)
        {
            if (chkUseGPU.Checked)
            {
                Program.useGPU = true;
                laMa = new LaMa(model_path);
            }
            else
            {
                Program.useGPU = false;
                laMa = new LaMa(model_path);
            }
        }
    }
}
 

using OpenCvSharp;
using System;
using System.Diagnostics;
using System.Drawing;
using System.Windows.Forms;

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

        string fileFilter = "*.*|*.bmp;*.jpg;*.jpeg;*.tiff;*.tiff;*.png";
        string image_path = "";
        string image_path_mask = "";
        string model_path;
        Mat image;
        Mat image_mask;
        LaMa laMa;

        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;
            pictureBox2.Image = null;
            textBox1.Text = "";
            Application.DoEvents();

            image = new Mat(image_path);
            image_mask = new Mat(image_path_mask);

            Stopwatch stopwatch = new Stopwatch();
            stopwatch.Start();

            Mat result = laMa.Run(image,image_mask);

            double costTime = stopwatch.Elapsed.TotalMilliseconds;

            if (chkUseGPU.Checked)
            {
                textBox1.Text = "GPU推理耗时:" + costTime + "ms";
            }
            else {
                textBox1.Text = "CPU推理耗时:" + costTime + "ms";
            }

            if (pictureBox2.Image!=null)
            {
                pictureBox2.Image.Dispose();
            }
            pictureBox2.Image = new Bitmap(result.ToMemoryStream());
            button2.Enabled = true;

            image_mask.Dispose();
            image.Dispose();
        }

        private void Form1_Load(object sender, EventArgs e)
        {
            model_path = "model/big_lama_regular_inpaint.onnx";
            laMa = new LaMa(model_path);

            image_path = "test_img/test.jpg";
            pictureBox1.Image = new Bitmap(image_path);

            image_path_mask = "test_img/mask.jpg";
            pictureBox3.Image = new Bitmap(image_path_mask);
        }

        private void chkUseGPU_CheckedChanged(object sender, EventArgs e)
        {
            if (chkUseGPU.Checked)
            {
                Program.useGPU = true;
                laMa = new LaMa(model_path);
            }
            else
            {
                Program.useGPU = false;
                laMa = new LaMa(model_path);
            }
        }
    }
}

Common.cs

using Microsoft.ML.OnnxRuntime;
using Microsoft.ML.OnnxRuntime.Tensors;
using OpenCvSharp;
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Threading.Tasks;

namespace Onnx_Demo
{
    internal class Common
    {
        public static void Preprocess(Mat image, Mat image_mask,  Tensor<float> input_tensor, Tensor<float> input_tensor_mask)
        {
            Cv2.Resize(image, image, new OpenCvSharp.Size(1504, 1000));
            // 输入Tensor
            for (int y = 0; y < image.Height; y++)
            {
                for (int x = 0; x < image.Width; x++)
                {
                    input_tensor[0, 0, y, x] = image.At<Vec3b>(y, x)[0] / 255.0f;
                    input_tensor[0, 1, y, x] = image.At<Vec3b>(y, x)[1] / 255.0f;
                    input_tensor[0, 2, y, x] = image.At<Vec3b>(y, x)[2] / 255.0f;
                }
            }

            Cv2.Resize(image_mask, image_mask, new OpenCvSharp.Size(1504, 1000));
            //膨胀核函数
            Mat element1 = new Mat();
            OpenCvSharp.Size size1 = new OpenCvSharp.Size(11, 11);
            element1 = Cv2.GetStructuringElement(MorphShapes.Rect, size1);

            //膨胀一次，让轮廓突出
            Mat dilation = new Mat();
            Cv2.Dilate(image_mask, image_mask, element1);

            //输入Tensor
            for (int y = 0; y < image_mask.Height; y++)
            {
                for (int x = 0; x < image_mask.Width; x++)
                {
                    float v = image_mask.At<Vec3b>(y, x)[0];
                    if (v > 127)
                    {
                        input_tensor_mask[0, 0, y, x] = 1.0f;
                    }
                    else
                    {
                        input_tensor_mask[0, 0, y, x] = 0.0f;
                    }
                }
            }

        }

        public static Mat Postprocess(IDisposableReadOnlyCollection<DisposableNamedOnnxValue> result_infer)
        {
            // 将输出结果转为DisposableNamedOnnxValue数组
            DisposableNamedOnnxValue[] results_onnxvalue = result_infer.ToArray();

            // 读取第一个节点输出并转为Tensor数据
            Tensor<float> result_tensors = results_onnxvalue[0].AsTensor<float>();

            float[] result_array = result_tensors.ToArray();

            for (int i = 0; i < result_array.Length; i++)
            {
                result_array[i] = Math.Max(0, Math.Min(255, result_array[i]));
            }

            Mat result = new Mat(1000, 1504, MatType.CV_32FC3, result_array);

            return result;
        }
    }
}

下载

源码下载