nancy ho 合集:Detecting hand gestures using Haarcascades training

来源：百度文库编辑：偶看新闻时间：2024/04/29 06:03:05

Detecting hand gestures using Haarcascades training

Haarcascades training (haartraining) is seemed an quick tool to achieve accurate hand gesture detection and recognition. The face and body detection examples included in openCV’s installation example folders (\opencv\data\haarcascades\) demonstrate how fast the haarcascades files help to do the job. More information about how to train the haarcascades files can go to sonots.com.

Many face-image databases have been provided for haarcascades training, e.g. http://www.face-rec.org/databases/. But hand images are less contributed for haarcascades training. At least, these are more difficult than face images to find in the internet. I was also asked by many blog readers for the haarcascades training file to support their hand detection or recognition related projects.

So I post an example of using haarcascades training file 1256617233-1-haarcascade_hand.xml for hand gesture detection.

The source codes can be downloaded from (or go to the download page):

http://download.andol.info/1256617233-1-haarcascade_hand.xml
http://download.andol.info/1256617233-2-haarcascade-hand.xml
http://download.andol.info/haarcascades-based%20hand%20detection.cpp

POSTS MAY BE OF INTEREST

Hand gesture recognition using Adaboost with SIFT
Transferring openCV from C++ to Java
A method of detecting and recognising hand gestures using openCV
Hand gesture recognition for HCI
A review of openCV of detection and recognition
Hand detection by using cvSnakeImage
“Secret Powers”: A finger detection demonstration
Hand detection and image recognition
Tracking a hand manipulating objects
Approaches for hands detection using openCV
Coming soon: A review of openCV detection & recognition
Pattern recognition and applications
5 inspiring hands detection videos
Hand detection using openCV

10 Responses to “Detecting hand gestures using Haarcascades training”

Jerry says: 4/ December/ 2011 at 4:28 PM
hai andol. may i ask u what software you use to run this program? i have use Microsoft Visual Studio 2010 Expres. and i try put in c++ but it failed to build because have so many eror… thanks ^^
Reply
- Andol says: 4/ December/ 2011 at 4:56 PM
  @Jerry
  the source code is based on C++, the application development environment is visual studio 2008, and the openCV version is 2.1. Please be aware of the differences of openCV versions’ setting ups in visual studio, as they have different head files and including libraries.
  Reply
  - jerry says: 12/ December/ 2011 at 8:39 AM
    @andol i also doing my project with c# @visual studio 2010. i have download your 1256617233-1-haarcascade_hand.xml. but it did’nt work @my project. would u help me to find out why it did’nt work. because for the other detection like face or nose my source code is working…this is my part project
    using System;
    using System.Collections.Generic;
    using System.Diagnostics;
    using System.ComponentModel;
    using System.Data;
    using System.Drawing;
    using System.Runtime.InteropServices;
    using System.Linq;
    using System.Text;
    using System.Windows.Forms;
    using Emgu.CV;
    using Emgu.CV.Structure;
    using Emgu.CV.UI;
    using Emgu.CV.GPU;
    namespace WindowsFormsApplication1
    {
    public partial class Form1 : Form
    {
    public Form1()
    {
    InitializeComponent();
    }
    public DialogResult res { get; set; }
    public Bitmap image { get; set; }
    public String path;
    private void Form1_Load(object sender, EventArgs e)
    {
    }
    private void pictureBox1_Click(object sender, EventArgs e)
    {
    }
    private void button1_Click_1(object sender, EventArgs e)
    {
    DialogResult res = openFileDialog1.ShowDialog();
    if (res == DialogResult.OK)
    {
    path = openFileDialog1.FileName;
    image = (Bitmap)Bitmap.FromFile(openFileDialog1.FileName);
    pictureBox1.Image = image;
    }
    }
    private void detect_Click(object sender, EventArgs e)
    {
    Image image = new Image(path); //Read the files as an 8-bit Bgr image
    Stopwatch watch;
    String bodyFileName = “full.xml”;
    String ubodyFileName = “haarcascade_mcs_upperbody.xml”;
    String lbodyFileName = “low.xml”;
    String faceFileName = “haarcascade_frontalface_alt.xml”;
    String eyeFileName = “haarcascade_eye.xml”;
    String noseFileName = “haarcascade_mcs_nose.xml”;
    String mouthFIleName = “mouth.xml”;
    String rearFileName = “haarcascade_mcs_leftear.xml”;
    String learFileName = “haarcascade_mcs_rightear.xml”;
    String handFileName = “hand.xml”;
    switch (cmbpilih.Text)
    {
    case “Full Body Detection”:
    if (GpuInvoke.HasCuda)
    {
    using (GpuCascadeClassifier body = new GpuCascadeClassifier(bodyFileName))
    {
    watch = Stopwatch.StartNew();
    using (GpuImage gpuImage = new GpuImage(image))
    using (GpuImage gpuGray = gpuImage.Convert())
    {
    Rectangle[] bodyRegion = body.DetectMultiScale(gpuGray, 1.1, 10, Size.Empty);
    foreach (Rectangle f in bodyRegion)
    {
    //draw the face detected in the 0th (gray) channel with blue color
    image.Draw(f, new Bgr(Color.Blue), 2);
    using (GpuImage bodyImg = gpuGray.GetSubRect(f))
    {
    //For some reason a clone is required.
    //Might be a bug of GpuCascadeClassifier in opencv
    using (GpuImage clone = bodyImg.Clone())
    {
    }
    }
    }
    }
    watch.Stop();
    }
    }
    else
    {
    //Read the HaarCascade objects
    using (HaarCascade body = new HaarCascade(bodyFileName))
    {
    watch = Stopwatch.StartNew();
    using (Image gray = image.Convert()) //Convert it to Grayscale
    {
    //normalizes brightness and increases contrast of the image
    gray._EqualizeHist();
    //Detect the faces from the gray scale image and store the locations as rectangle
    //The first dimensional is the channel
    //The second dimension is the index of the rectangle in the specific channel
    MCvAvgComp[] bodysDetected = body.Detect(
    gray,
    1.1,
    10,
    Emgu.CV.CvEnum.HAAR_DETECTION_TYPE.DO_CANNY_PRUNING,
    new Size(20, 20));
    foreach (MCvAvgComp f in bodysDetected)
    {
    //draw the face detected in the 0th (gray) channel with blue color
    image.Draw(f.rect, new Bgr(Color.Blue), 2);
    //Set the region of interest on the faces
    gray.ROI = f.rect;
    }
    }
    watch.Stop();
    }
    }
    pictureBox1.Image = image.ToBitmap();
    //display the image
    //ImageViewer.Show(image);// String.Format(
    //”Completed face and eye detection using {0} in {1} milliseconds”,
    //GpuInvoke.HasCuda ? “GPU” : “CPU”,
    //watch.ElapsedMilliseconds));
    break;
    case “Upper Body Detection”:
    if (GpuInvoke.HasCuda)
    {
    using (GpuCascadeClassifier ubody = new GpuCascadeClassifier(ubodyFileName))
    {
    watch = Stopwatch.StartNew();
    using (GpuImage gpuImage = new GpuImage(image))
    using (GpuImage gpuGray = gpuImage.Convert())
    {
    Rectangle[] ubodyRegion = ubody.DetectMultiScale(gpuGray, 1.1, 10, Size.Empty);
    foreach (Rectangle f in ubodyRegion)
    {
    //draw the face detected in the 0th (gray) channel with blue color
    image.Draw(f, new Bgr(Color.Blue), 2);
    using (GpuImage ubodyImg = gpuGray.GetSubRect(f))
    {
    //For some reason a clone is required.
    //Might be a bug of GpuCascadeClassifier in opencv
    using (GpuImage clone = ubodyImg.Clone())
    {
    }
    }
    }
    }
    watch.Stop();
    }
    }
    else
    {
    //Read the HaarCascade objects
    using (HaarCascade ubody = new HaarCascade(ubodyFileName))
    {
    watch = Stopwatch.StartNew();
    using (Image gray = image.Convert()) //Convert it to Grayscale
    {
    //normalizes brightness and increases contrast of the image
    gray._EqualizeHist();
    //Detect the faces from the gray scale image and store the locations as rectangle
    //The first dimensional is the channel
    //The second dimension is the index of the rectangle in the specific channel
    MCvAvgComp[] ubodysDetected = ubody.Detect(
    gray,
    1.1,
    10,
    Emgu.CV.CvEnum.HAAR_DETECTION_TYPE.DO_CANNY_PRUNING,
    new Size(20, 20));
    foreach (MCvAvgComp f in ubodysDetected)
    {
    //draw the face detected in the 0th (gray) channel with blue color
    image.Draw(f.rect, new Bgr(Color.Blue), 2);
    //Set the region of interest on the faces
    gray.ROI = f.rect;
    }
    }
    watch.Stop();
    }
    }
    pictureBox1.Image = image.ToBitmap();
    break;
    case “Lower Body Detection”:
    if (GpuInvoke.HasCuda)
    {
    using (GpuCascadeClassifier lbody = new GpuCascadeClassifier(lbodyFileName))
    {
    watch = Stopwatch.StartNew();
    using (GpuImage gpuImage = new GpuImage(image))
    using (GpuImage gpuGray = gpuImage.Convert())
    {
    Rectangle[] lbodyRegion = lbody.DetectMultiScale(gpuGray, 1.1, 10, Size.Empty);
    foreach (Rectangle f in lbodyRegion)
    {
    //draw the face detected in the 0th (gray) channel with blue color
    image.Draw(f, new Bgr(Color.Blue), 2);
    using (GpuImage lbodyImg = gpuGray.GetSubRect(f))
    {
    //For some reason a clone is required.
    //Might be a bug of GpuCascadeClassifier in opencv
    using (GpuImage clone = lbodyImg.Clone())
    {
    }
    }
    }
    }
    watch.Stop();
    }
    }
    else
    {
    //Read the HaarCascade objects
    using (HaarCascade lbody = new HaarCascade(lbodyFileName))
    {
    watch = Stopwatch.StartNew();
    using (Image gray = image.Convert()) //Convert it to Grayscale
    {
    //normalizes brightness and increases contrast of the image
    gray._EqualizeHist();
    //Detect the faces from the gray scale image and store the locations as rectangle
    //The first dimensional is the channel
    //The second dimension is the index of the rectangle in the specific channel
    MCvAvgComp[] lbodysDetected = lbody.Detect(
    gray,
    1.1,
    10,
    Emgu.CV.CvEnum.HAAR_DETECTION_TYPE.DO_CANNY_PRUNING,
    new Size(20, 20));
    foreach (MCvAvgComp f in lbodysDetected)
    {
    //draw the face detected in the 0th (gray) channel with blue color
    image.Draw(f.rect, new Bgr(Color.Blue), 2);
    //Set the region of interest on the faces
    gray.ROI = f.rect;
    }
    }
    watch.Stop();
    }
    }
    pictureBox1.Image = image.ToBitmap();
    break;
    case “Face Detection”:
    if (GpuInvoke.HasCuda)
    {
    using (GpuCascadeClassifier face = new GpuCascadeClassifier(faceFileName))
    {
    watch = Stopwatch.StartNew();
    using (GpuImage gpuImage = new GpuImage(image))
    using (GpuImage gpuGray = gpuImage.Convert())
    {
    Rectangle[] faceRegion = face.DetectMultiScale(gpuGray, 1.1, 10, Size.Empty);
    foreach (Rectangle f in faceRegion)
    {
    //draw the face detected in the 0th (gray) channel with blue color
    image.Draw(f, new Bgr(Color.Blue), 2);
    using (GpuImage faceImg = gpuGray.GetSubRect(f))
    {
    //For some reason a clone is required.
    //Might be a bug of GpuCascadeClassifier in opencv
    using (GpuImage clone = faceImg.Clone())
    {
    }
    }
    }
    }
    watch.Stop();
    }
    }
    else
    {
    //Read the HaarCascade objects
    using (HaarCascade face = new HaarCascade(faceFileName))
    {
    watch = Stopwatch.StartNew();
    using (Image gray = image.Convert()) //Convert it to Grayscale
    {
    //normalizes brightness and increases contrast of the image
    gray._EqualizeHist();
    //Detect the faces from the gray scale image and store the locations as rectangle
    //The first dimensional is the channel
    //The second dimension is the index of the rectangle in the specific channel
    MCvAvgComp[] facesDetected = face.Detect(
    gray,
    1.1,
    10,
    Emgu.CV.CvEnum.HAAR_DETECTION_TYPE.DO_CANNY_PRUNING,
    new Size(20, 20));
    foreach (MCvAvgComp f in facesDetected)
    {
    //draw the face detected in the 0th (gray) channel with blue color
    image.Draw(f.rect, new Bgr(Color.Blue), 2);
    //Set the region of interest on the faces
    gray.ROI = f.rect;
    }
    }
    watch.Stop();
    }
    }
    pictureBox1.Image = image.ToBitmap();
    break;
    case “Eye Detection”:
    if (GpuInvoke.HasCuda)
    {
    using (GpuCascadeClassifier eye = new GpuCascadeClassifier(eyeFileName))
    {
    watch = Stopwatch.StartNew();
    using (GpuImage gpuImage = new GpuImage(image))
    using (GpuImage gpuGray = gpuImage.Convert())
    {
    Rectangle[] eyeRegion = eye.DetectMultiScale(gpuGray, 1.1, 10, Size.Empty);
    foreach (Rectangle f in eyeRegion)
    {
    //draw the face detected in the 0th (gray) channel with blue color
    image.Draw(f, new Bgr(Color.Blue), 2);
    using (GpuImage eyeImg = gpuGray.GetSubRect(f))
    {
    //For some reason a clone is required.
    //Might be a bug of GpuCascadeClassifier in opencv
    using (GpuImage clone = eyeImg.Clone())
    {
    }
    }
    }
    }
    watch.Stop();
    }
    }
    else
    {
    //Read the HaarCascade objects
    using (HaarCascade eye = new HaarCascade(eyeFileName))
    {
    watch = Stopwatch.StartNew();
    using (Image gray = image.Convert()) //Convert it to Grayscale
    {
    //normalizes brightness and increases contrast of the image
    gray._EqualizeHist();
    //Detect the faces from the gray scale image and store the locations as rectangle
    //The first dimensional is the channel
    //The second dimension is the index of the rectangle in the specific channel
    MCvAvgComp[] eyesDetected = eye.Detect(
    gray,
    1.1,
    10,
    Emgu.CV.CvEnum.HAAR_DETECTION_TYPE.DO_CANNY_PRUNING,
    new Size(20, 20));
    foreach (MCvAvgComp f in eyesDetected)
    {
    //draw the face detected in the 0th (gray) channel with blue color
    image.Draw(f.rect, new Bgr(Color.Blue), 2);
    //Set the region of interest on the faces
    gray.ROI = f.rect;
    }
    }
    watch.Stop();
    }
    }
    pictureBox1.Image = image.ToBitmap();
    break;
    case “Nose Detection”:
    if (GpuInvoke.HasCuda)
    {
    using (GpuCascadeClassifier nose = new GpuCascadeClassifier(noseFileName))
    {
    watch = Stopwatch.StartNew();
    using (GpuImage gpuImage = new GpuImage(image))
    using (GpuImage gpuGray = gpuImage.Convert())
    {
    Rectangle[] noseRegion = nose.DetectMultiScale(gpuGray, 1.1, 10, Size.Empty);
    foreach (Rectangle f in noseRegion)
    {
    //draw the face detected in the 0th (gray) channel with blue color
    image.Draw(f, new Bgr(Color.Blue), 2);
    using (GpuImage noseImg = gpuGray.GetSubRect(f))
    {
    //For some reason a clone is required.
    //Might be a bug of GpuCascadeClassifier in opencv
    using (GpuImage clone = noseImg.Clone())
    {
    }
    }
    }
    }
    watch.Stop();
    }
    }
    else
    {
    //Read the HaarCascade objects
    using (HaarCascade nose = new HaarCascade(noseFileName))
    {
    watch = Stopwatch.StartNew();
    using (Image gray = image.Convert()) //Convert it to Grayscale
    {
    //normalizes brightness and increases contrast of the image
    gray._EqualizeHist();
    //Detect the faces from the gray scale image and store the locations as rectangle
    //The first dimensional is the channel
    //The second dimension is the index of the rectangle in the specific channel
    MCvAvgComp[] nosesDetected = nose.Detect(
    gray,
    1.1,
    10,
    Emgu.CV.CvEnum.HAAR_DETECTION_TYPE.DO_CANNY_PRUNING,
    new Size(20, 20));
    foreach (MCvAvgComp f in nosesDetected)
    {
    //draw the face detected in the 0th (gray) channel with blue color
    image.Draw(f.rect, new Bgr(Color.Blue), 2);
    //Set the region of interest on the faces
    gray.ROI = f.rect;
    }
    }
    watch.Stop();
    }
    }
    pictureBox1.Image = image.ToBitmap();
    break;
    case “Mouth Detection”:
    if (GpuInvoke.HasCuda)
    {
    using (GpuCascadeClassifier mouth = new GpuCascadeClassifier(mouthFIleName))
    {
    watch = Stopwatch.StartNew();
    using (GpuImage gpuImage = new GpuImage(image))
    using (GpuImage gpuGray = gpuImage.Convert())
    {
    Rectangle[] mouthRegion = mouth.DetectMultiScale(gpuGray, 1.1, 10, Size.Empty);
    foreach (Rectangle f in mouthRegion)
    {
    //draw the face detected in the 0th (gray) channel with blue color
    image.Draw(f, new Bgr(Color.Blue), 2);
    using (GpuImage mouthImg = gpuGray.GetSubRect(f))
    {
    //For some reason a clone is required.
    //Might be a bug of GpuCascadeClassifier in opencv
    using (GpuImage clone = mouthImg.Clone())
    {
    }
    }
    }
    }
    watch.Stop();
    }
    }
    else
    {
    //Read the HaarCascade objects
    using (HaarCascade mouth = new HaarCascade(mouthFIleName))
    {
    watch = Stopwatch.StartNew();
    using (Image gray = image.Convert()) //Convert it to Grayscale
    {
    //normalizes brightness and increases contrast of the image
    gray._EqualizeHist();
    //Detect the faces from the gray scale image and store the locations as rectangle
    //The first dimensional is the channel
    //The second dimension is the index of the rectangle in the specific channel
    MCvAvgComp[] mouthsDetected = mouth.Detect(
    gray,
    1.1,
    10,
    Emgu.CV.CvEnum.HAAR_DETECTION_TYPE.DO_CANNY_PRUNING,
    new Size(20, 20));
    foreach (MCvAvgComp f in mouthsDetected)
    {
    //draw the face detected in the 0th (gray) channel with blue color
    image.Draw(f.rect, new Bgr(Color.Blue), 2);
    //Set the region of interest on the faces
    gray.ROI = f.rect;
    }
    }
    watch.Stop();
    }
    }
    pictureBox1.Image = image.ToBitmap();
    break;
    case “Right Ear Detection”:
    if (GpuInvoke.HasCuda)
    {
    using (GpuCascadeClassifier rear = new GpuCascadeClassifier(rearFileName))
    {
    watch = Stopwatch.StartNew();
    using (GpuImage gpuImage = new GpuImage(image))
    using (GpuImage gpuGray = gpuImage.Convert())
    {
    Rectangle[] rearRegion = rear.DetectMultiScale(gpuGray, 1.1, 10, Size.Empty);
    foreach (Rectangle f in rearRegion)
    {
    //draw the face detected in the 0th (gray) channel with blue color
    image.Draw(f, new Bgr(Color.Blue), 2);
    using (GpuImage rearImg = gpuGray.GetSubRect(f))
    {
    //For some reason a clone is required.
    //Might be a bug of GpuCascadeClassifier in opencv
    using (GpuImage clone = rearImg.Clone())
    {
    }
    }
    }
    }
    watch.Stop();
    }
    }
    else
    {
    //Read the HaarCascade objects
    using (HaarCascade rear = new HaarCascade(rearFileName))
    {
    watch = Stopwatch.StartNew();
    using (Image gray = image.Convert()) //Convert it to Grayscale
    {
    //normalizes brightness and increases contrast of the image
    gray._EqualizeHist();
    //Detect the faces from the gray scale image and store the locations as rectangle
    //The first dimensional is the channel
    //The second dimension is the index of the rectangle in the specific channel
    MCvAvgComp[] rearsDetected = rear.Detect(
    gray,
    1.1,
    10,
    Emgu.CV.CvEnum.HAAR_DETECTION_TYPE.DO_CANNY_PRUNING,
    new Size(20, 20));
    foreach (MCvAvgComp f in rearsDetected)
    {
    //draw the face detected in the 0th (gray) channel with blue color
    image.Draw(f.rect, new Bgr(Color.Blue), 2);
    //Set the region of interest on the faces
    gray.ROI = f.rect;
    }
    }
    watch.Stop();
    }
    }
    pictureBox1.Image = image.ToBitmap();
    break;
    case “Left Ear Detection”:
    if (GpuInvoke.HasCuda)
    {
    using (GpuCascadeClassifier lear = new GpuCascadeClassifier(learFileName))
    {
    watch = Stopwatch.StartNew();
    using (GpuImage gpuImage = new GpuImage(image))
    using (GpuImage gpuGray = gpuImage.Convert())
    {
    Rectangle[] learRegion = lear.DetectMultiScale(gpuGray, 1.1, 10, Size.Empty);
    foreach (Rectangle f in learRegion)
    {
    //draw the face detected in the 0th (gray) channel with blue color
    image.Draw(f, new Bgr(Color.Blue), 2);
    using (GpuImage learImg = gpuGray.GetSubRect(f))
    {
    //For some reason a clone is required.
    //Might be a bug of GpuCascadeClassifier in opencv
    using (GpuImage clone = learImg.Clone())
    {
    }
    }
    }
    }
    watch.Stop();
    }
    }
    else
    {
    //Read the HaarCascade objects
    using (HaarCascade lear = new HaarCascade(learFileName))
    {
    watch = Stopwatch.StartNew();
    using (Image gray = image.Convert()) //Convert it to Grayscale
    {
    //normalizes brightness and increases contrast of the image
    gray._EqualizeHist();
    //Detect the faces from the gray scale image and store the locations as rectangle
    //The first dimensional is the channel
    //The second dimension is the index of the rectangle in the specific channel
    MCvAvgComp[] learsDetected = lear.Detect(
    gray,
    1.1,
    10,
    Emgu.CV.CvEnum.HAAR_DETECTION_TYPE.DO_CANNY_PRUNING,
    new Size(20, 20));
    foreach (MCvAvgComp f in learsDetected)
    {
    //draw the face detected in the 0th (gray) channel with blue color
    image.Draw(f.rect, new Bgr(Color.Blue), 2);
    //Set the region of interest on the faces
    gray.ROI = f.rect;
    }
    }
    watch.Stop();
    }
    }
    pictureBox1.Image = image.ToBitmap();
    break;
    case “Hand Detection”:
    if (GpuInvoke.HasCuda)
    {
    using (GpuCascadeClassifier hand = new GpuCascadeClassifier(handFileName))
    {
    watch = Stopwatch.StartNew();
    using (GpuImage gpuImage = new GpuImage(image))
    using (GpuImage gpuGray = gpuImage.Convert())
    {
    Rectangle[] handRegion = hand.DetectMultiScale(gpuGray, 1.1, 10, Size.Empty);
    foreach (Rectangle f in handRegion)
    {
    //draw the face detected in the 0th (gray) channel with blue color
    image.Draw(f, new Bgr(Color.Blue), 2);
    using (GpuImage faceImg = gpuGray.GetSubRect(f))
    {
    //For some reason a clone is required.
    //Might be a bug of GpuCascadeClassifier in opencv
    using (GpuImage clone = faceImg.Clone())
    {
    }
    }
    }
    }
    watch.Stop();
    }
    }
    else
    {
    //Read the HaarCascade objects
    using (HaarCascade hand = new HaarCascade(handFileName))
    {
    watch = Stopwatch.StartNew();
    using (Image gray = image.Convert()) //Convert it to Grayscale
    {
    //normalizes brightness and increases contrast of the image
    gray._EqualizeHist();
    //Detect the faces from the gray scale image and store the locations as rectangle
    //The first dimensional is the channel
    //The second dimension is the index of the rectangle in the specific channel
    MCvAvgComp[] handsDetected = hand.Detect(
    gray,
    1.1,
    10,
    Emgu.CV.CvEnum.HAAR_DETECTION_TYPE.DO_CANNY_PRUNING,
    new Size(20, 20));
    foreach (MCvAvgComp f in handsDetected)
    {
    //draw the face detected in the 0th (gray) channel with blue color
    image.Draw(f.rect, new Bgr(Color.Blue), 2);
    //Set the region of interest on the faces
    gray.ROI = f.rect;
    }
    }
    watch.Stop();
    }
    }
    pictureBox1.Image = image.ToBitmap();
    break;
    }
    }
    }
    }
    Reply
链接大放送 | 增强视觉 | 计算机视觉增强现实 says: 7/ December/ 2011 at 1:42 PM
[...] 来自Andol的hand detector。 [...]
Reply
Thought this was cool: 链接大放送 ? CWYAlpha says: 7/ December/ 2011 at 3:09 PM
[...] 来自Andol的hand detector。 [...]
Reply
jerry says: 12/ December/ 2011 at 8:43 AM
hai andol i have doing my final project for human body detection with c# @visual studio 2010, with emguCV(openCV in c#), i have download your haarcascade for hand, but it not work at my project, but for the other detection like nose,face,eye with the haarcascade too, my source code is working.. would you give me some sugest for it? this is my source code
using System;
using System.Collections.Generic;
using System.Diagnostics;
using System.ComponentModel;
using System.Data;
using System.Drawing;
using System.Runtime.InteropServices;
using System.Linq;
using System.Text;
using System.Windows.Forms;
using Emgu.CV;
using Emgu.CV.Structure;
using Emgu.CV.UI;
using Emgu.CV.GPU;
namespace WindowsFormsApplication1
{
public partial class Form1 : Form
{
public Form1()
{
InitializeComponent();
}
public DialogResult res { get; set; }
public Bitmap image { get; set; }
public String path;
private void Form1_Load(object sender, EventArgs e)
{
}
private void pictureBox1_Click(object sender, EventArgs e)
{
}
private void button1_Click_1(object sender, EventArgs e)
{
DialogResult res = openFileDialog1.ShowDialog();
if (res == DialogResult.OK)
{
path = openFileDialog1.FileName;
image = (Bitmap)Bitmap.FromFile(openFileDialog1.FileName);
pictureBox1.Image = image;
}
}
private void detect_Click(object sender, EventArgs e)
{
Image image = new Image(path); //Read the files as an 8-bit Bgr image
Stopwatch watch;
String bodyFileName = “full.xml”;
String ubodyFileName = “haarcascade_mcs_upperbody.xml”;
String lbodyFileName = “low.xml”;
String faceFileName = “haarcascade_frontalface_alt.xml”;
String eyeFileName = “haarcascade_eye.xml”;
String noseFileName = “haarcascade_mcs_nose.xml”;
String mouthFIleName = “mouth.xml”;
String rearFileName = “haarcascade_mcs_leftear.xml”;
String learFileName = “haarcascade_mcs_rightear.xml”;
String handFileName = “hand.xml”;
switch (cmbpilih.Text)
{
case “Full Body Detection”:
if (GpuInvoke.HasCuda)
{
using (GpuCascadeClassifier body = new GpuCascadeClassifier(bodyFileName))
{
watch = Stopwatch.StartNew();
using (GpuImage gpuImage = new GpuImage(image))
using (GpuImage gpuGray = gpuImage.Convert())
{
Rectangle[] bodyRegion = body.DetectMultiScale(gpuGray, 1.1, 10, Size.Empty);
foreach (Rectangle f in bodyRegion)
{
//draw the face detected in the 0th (gray) channel with blue color
image.Draw(f, new Bgr(Color.Blue), 2);
using (GpuImage bodyImg = gpuGray.GetSubRect(f))
{
//For some reason a clone is required.
//Might be a bug of GpuCascadeClassifier in opencv
using (GpuImage clone = bodyImg.Clone())
{
}
}
}
}
watch.Stop();
}
}
else
{
//Read the HaarCascade objects
using (HaarCascade body = new HaarCascade(bodyFileName))
{
watch = Stopwatch.StartNew();
using (Image gray = image.Convert()) //Convert it to Grayscale
{
//normalizes brightness and increases contrast of the image
gray._EqualizeHist();
//Detect the faces from the gray scale image and store the locations as rectangle
//The first dimensional is the channel
//The second dimension is the index of the rectangle in the specific channel
MCvAvgComp[] bodysDetected = body.Detect(
gray,
1.1,
10,
Emgu.CV.CvEnum.HAAR_DETECTION_TYPE.DO_CANNY_PRUNING,
new Size(20, 20));
foreach (MCvAvgComp f in bodysDetected)
{
//draw the face detected in the 0th (gray) channel with blue color
image.Draw(f.rect, new Bgr(Color.Blue), 2);
//Set the region of interest on the faces
gray.ROI = f.rect;
}
}
watch.Stop();
}
}
pictureBox1.Image = image.ToBitmap();
//display the image
//ImageViewer.Show(image);// String.Format(
//”Completed face and eye detection using {0} in {1} milliseconds”,
//GpuInvoke.HasCuda ? “GPU” : “CPU”,
//watch.ElapsedMilliseconds));
break;
case “Upper Body Detection”:
if (GpuInvoke.HasCuda)
{
using (GpuCascadeClassifier ubody = new GpuCascadeClassifier(ubodyFileName))
{
watch = Stopwatch.StartNew();
using (GpuImage gpuImage = new GpuImage(image))
using (GpuImage gpuGray = gpuImage.Convert())
{
Rectangle[] ubodyRegion = ubody.DetectMultiScale(gpuGray, 1.1, 10, Size.Empty);
foreach (Rectangle f in ubodyRegion)
{
//draw the face detected in the 0th (gray) channel with blue color
image.Draw(f, new Bgr(Color.Blue), 2);
using (GpuImage ubodyImg = gpuGray.GetSubRect(f))
{
//For some reason a clone is required.
//Might be a bug of GpuCascadeClassifier in opencv
using (GpuImage clone = ubodyImg.Clone())
{
}
}
}
}
watch.Stop();
}
}
else
{
//Read the HaarCascade objects
using (HaarCascade ubody = new HaarCascade(ubodyFileName))
{
watch = Stopwatch.StartNew();
using (Image gray = image.Convert()) //Convert it to Grayscale
{
//normalizes brightness and increases contrast of the image
gray._EqualizeHist();
//Detect the faces from the gray scale image and store the locations as rectangle
//The first dimensional is the channel
//The second dimension is the index of the rectangle in the specific channel
MCvAvgComp[] ubodysDetected = ubody.Detect(
gray,
1.1,
10,
Emgu.CV.CvEnum.HAAR_DETECTION_TYPE.DO_CANNY_PRUNING,
new Size(20, 20));
foreach (MCvAvgComp f in ubodysDetected)
{
//draw the face detected in the 0th (gray) channel with blue color
image.Draw(f.rect, new Bgr(Color.Blue), 2);
//Set the region of interest on the faces
gray.ROI = f.rect;
}
}
watch.Stop();
}
}
pictureBox1.Image = image.ToBitmap();
break;
case “Lower Body Detection”:
if (GpuInvoke.HasCuda)
{
using (GpuCascadeClassifier lbody = new GpuCascadeClassifier(lbodyFileName))
{
watch = Stopwatch.StartNew();
using (GpuImage gpuImage = new GpuImage(image))
using (GpuImage gpuGray = gpuImage.Convert())
{
Rectangle[] lbodyRegion = lbody.DetectMultiScale(gpuGray, 1.1, 10, Size.Empty);
foreach (Rectangle f in lbodyRegion)
{
//draw the face detected in the 0th (gray) channel with blue color
image.Draw(f, new Bgr(Color.Blue), 2);
using (GpuImage lbodyImg = gpuGray.GetSubRect(f))
{
//For some reason a clone is required.
//Might be a bug of GpuCascadeClassifier in opencv
using (GpuImage clone = lbodyImg.Clone())
{
}
}
}
}
watch.Stop();
}
}
else
{
//Read the HaarCascade objects
using (HaarCascade lbody = new HaarCascade(lbodyFileName))
{
watch = Stopwatch.StartNew();
using (Image gray = image.Convert()) //Convert it to Grayscale
{
//normalizes brightness and increases contrast of the image
gray._EqualizeHist();
//Detect the faces from the gray scale image and store the locations as rectangle
//The first dimensional is the channel
//The second dimension is the index of the rectangle in the specific channel
MCvAvgComp[] lbodysDetected = lbody.Detect(
gray,
1.1,
10,
Emgu.CV.CvEnum.HAAR_DETECTION_TYPE.DO_CANNY_PRUNING,
new Size(20, 20));
foreach (MCvAvgComp f in lbodysDetected)
{
//draw the face detected in the 0th (gray) channel with blue color
image.Draw(f.rect, new Bgr(Color.Blue), 2);
//Set the region of interest on the faces
gray.ROI = f.rect;
}
}
watch.Stop();
}
}
pictureBox1.Image = image.ToBitmap();
break;
case “Face Detection”:
if (GpuInvoke.HasCuda)
{
using (GpuCascadeClassifier face = new GpuCascadeClassifier(faceFileName))
{
watch = Stopwatch.StartNew();
using (GpuImage gpuImage = new GpuImage(image))
using (GpuImage gpuGray = gpuImage.Convert())
{
Rectangle[] faceRegion = face.DetectMultiScale(gpuGray, 1.1, 10, Size.Empty);
foreach (Rectangle f in faceRegion)
{
//draw the face detected in the 0th (gray) channel with blue color
image.Draw(f, new Bgr(Color.Blue), 2);
using (GpuImage faceImg = gpuGray.GetSubRect(f))
{
//For some reason a clone is required.
//Might be a bug of GpuCascadeClassifier in opencv
using (GpuImage clone = faceImg.Clone())
{
}
}
}
}
watch.Stop();
}
}
else
{
//Read the HaarCascade objects
using (HaarCascade face = new HaarCascade(faceFileName))
{
watch = Stopwatch.StartNew();
using (Image gray = image.Convert()) //Convert it to Grayscale
{
//normalizes brightness and increases contrast of the image
gray._EqualizeHist();
//Detect the faces from the gray scale image and store the locations as rectangle
//The first dimensional is the channel
//The second dimension is the index of the rectangle in the specific channel
MCvAvgComp[] facesDetected = face.Detect(
gray,
1.1,
10,
Emgu.CV.CvEnum.HAAR_DETECTION_TYPE.DO_CANNY_PRUNING,
new Size(20, 20));
foreach (MCvAvgComp f in facesDetected)
{
//draw the face detected in the 0th (gray) channel with blue color
image.Draw(f.rect, new Bgr(Color.Blue), 2);
//Set the region of interest on the faces
gray.ROI = f.rect;
}
}
watch.Stop();
}
}
pictureBox1.Image = image.ToBitmap();
break;
case “Eye Detection”:
if (GpuInvoke.HasCuda)
{
using (GpuCascadeClassifier eye = new GpuCascadeClassifier(eyeFileName))
{
watch = Stopwatch.StartNew();
using (GpuImage gpuImage = new GpuImage(image))
using (GpuImage gpuGray = gpuImage.Convert())
{
Rectangle[] eyeRegion = eye.DetectMultiScale(gpuGray, 1.1, 10, Size.Empty);
foreach (Rectangle f in eyeRegion)
{
//draw the face detected in the 0th (gray) channel with blue color
image.Draw(f, new Bgr(Color.Blue), 2);
using (GpuImage eyeImg = gpuGray.GetSubRect(f))
{
//For some reason a clone is required.
//Might be a bug of GpuCascadeClassifier in opencv
using (GpuImage clone = eyeImg.Clone())
{
}
}
}
}
watch.Stop();
}
}
else
{
//Read the HaarCascade objects
using (HaarCascade eye = new HaarCascade(eyeFileName))
{
watch = Stopwatch.StartNew();
using (Image gray = image.Convert()) //Convert it to Grayscale
{
//normalizes brightness and increases contrast of the image
gray._EqualizeHist();
//Detect the faces from the gray scale image and store the locations as rectangle
//The first dimensional is the channel
//The second dimension is the index of the rectangle in the specific channel
MCvAvgComp[] eyesDetected = eye.Detect(
gray,
1.1,
10,
Emgu.CV.CvEnum.HAAR_DETECTION_TYPE.DO_CANNY_PRUNING,
new Size(20, 20));
foreach (MCvAvgComp f in eyesDetected)
{
//draw the face detected in the 0th (gray) channel with blue color
image.Draw(f.rect, new Bgr(Color.Blue), 2);
//Set the region of interest on the faces
gray.ROI = f.rect;
}
}
watch.Stop();
}
}
pictureBox1.Image = image.ToBitmap();
break;
case “Nose Detection”:
if (GpuInvoke.HasCuda)
{
using (GpuCascadeClassifier nose = new GpuCascadeClassifier(noseFileName))
{
watch = Stopwatch.StartNew();
using (GpuImage gpuImage = new GpuImage(image))
using (GpuImage gpuGray = gpuImage.Convert())
{
Rectangle[] noseRegion = nose.DetectMultiScale(gpuGray, 1.1, 10, Size.Empty);
foreach (Rectangle f in noseRegion)
{
//draw the face detected in the 0th (gray) channel with blue color
image.Draw(f, new Bgr(Color.Blue), 2);
using (GpuImage noseImg = gpuGray.GetSubRect(f))
{
//For some reason a clone is required.
//Might be a bug of GpuCascadeClassifier in opencv
using (GpuImage clone = noseImg.Clone())
{
}
}
}
}
watch.Stop();
}
}
else
{
//Read the HaarCascade objects
using (HaarCascade nose = new HaarCascade(noseFileName))
{
watch = Stopwatch.StartNew();
using (Image gray = image.Convert()) //Convert it to Grayscale
{
//normalizes brightness and increases contrast of the image
gray._EqualizeHist();
//Detect the faces from the gray scale image and store the locations as rectangle
//The first dimensional is the channel
//The second dimension is the index of the rectangle in the specific channel
MCvAvgComp[] nosesDetected = nose.Detect(
gray,
1.1,
10,
Emgu.CV.CvEnum.HAAR_DETECTION_TYPE.DO_CANNY_PRUNING,
new Size(20, 20));
foreach (MCvAvgComp f in nosesDetected)
{
//draw the face detected in the 0th (gray) channel with blue color
image.Draw(f.rect, new Bgr(Color.Blue), 2);
//Set the region of interest on the faces
gray.ROI = f.rect;
}
}
watch.Stop();
}
}
pictureBox1.Image = image.ToBitmap();
break;
case “Mouth Detection”:
if (GpuInvoke.HasCuda)
{
using (GpuCascadeClassifier mouth = new GpuCascadeClassifier(mouthFIleName))
{
watch = Stopwatch.StartNew();
using (GpuImage gpuImage = new GpuImage(image))
using (GpuImage gpuGray = gpuImage.Convert())
{
Rectangle[] mouthRegion = mouth.DetectMultiScale(gpuGray, 1.1, 10, Size.Empty);
foreach (Rectangle f in mouthRegion)
{
//draw the face detected in the 0th (gray) channel with blue color
image.Draw(f, new Bgr(Color.Blue), 2);
using (GpuImage mouthImg = gpuGray.GetSubRect(f))
{
//For some reason a clone is required.
//Might be a bug of GpuCascadeClassifier in opencv
using (GpuImage clone = mouthImg.Clone())
{
}
}
}
}
watch.Stop();
}
}
else
{
//Read the HaarCascade objects
using (HaarCascade mouth = new HaarCascade(mouthFIleName))
{
watch = Stopwatch.StartNew();
using (Image gray = image.Convert()) //Convert it to Grayscale
{
//normalizes brightness and increases contrast of the image
gray._EqualizeHist();
//Detect the faces from the gray scale image and store the locations as rectangle
//The first dimensional is the channel
//The second dimension is the index of the rectangle in the specific channel
MCvAvgComp[] mouthsDetected = mouth.Detect(
gray,
1.1,
10,
Emgu.CV.CvEnum.HAAR_DETECTION_TYPE.DO_CANNY_PRUNING,
new Size(20, 20));
foreach (MCvAvgComp f in mouthsDetected)
{
//draw the face detected in the 0th (gray) channel with blue color
image.Draw(f.rect, new Bgr(Color.Blue), 2);
//Set the region of interest on the faces
gray.ROI = f.rect;
}
}
watch.Stop();
}
}
pictureBox1.Image = image.ToBitmap();
break;
case “Right Ear Detection”:
if (GpuInvoke.HasCuda)
{
using (GpuCascadeClassifier rear = new GpuCascadeClassifier(rearFileName))
{
watch = Stopwatch.StartNew();
using (GpuImage gpuImage = new GpuImage(image))
using (GpuImage gpuGray = gpuImage.Convert())
{
Rectangle[] rearRegion = rear.DetectMultiScale(gpuGray, 1.1, 10, Size.Empty);
foreach (Rectangle f in rearRegion)
{
//draw the face detected in the 0th (gray) channel with blue color
image.Draw(f, new Bgr(Color.Blue), 2);
using (GpuImage rearImg = gpuGray.GetSubRect(f))
{
//For some reason a clone is required.
//Might be a bug of GpuCascadeClassifier in opencv
using (GpuImage clone = rearImg.Clone())
{
}
}
}
}
watch.Stop();
}
}
else
{
//Read the HaarCascade objects
using (HaarCascade rear = new HaarCascade(rearFileName))
{
watch = Stopwatch.StartNew();
using (Image gray = image.Convert()) //Convert it to Grayscale
{
//normalizes brightness and increases contrast of the image
gray._EqualizeHist();
//Detect the faces from the gray scale image and store the locations as rectangle
//The first dimensional is the channel
//The second dimension is the index of the rectangle in the specific channel
MCvAvgComp[] rearsDetected = rear.Detect(
gray,
1.1,
10,
Emgu.CV.CvEnum.HAAR_DETECTION_TYPE.DO_CANNY_PRUNING,
new Size(20, 20));
foreach (MCvAvgComp f in rearsDetected)
{
//draw the face detected in the 0th (gray) channel with blue color
image.Draw(f.rect, new Bgr(Color.Blue), 2);
//Set the region of interest on the faces
gray.ROI = f.rect;
}
}
watch.Stop();
}
}
pictureBox1.Image = image.ToBitmap();
break;
case “Left Ear Detection”:
if (GpuInvoke.HasCuda)
{
using (GpuCascadeClassifier lear = new GpuCascadeClassifier(learFileName))
{
watch = Stopwatch.StartNew();
using (GpuImage gpuImage = new GpuImage(image))
using (GpuImage gpuGray = gpuImage.Convert())
{
Rectangle[] learRegion = lear.DetectMultiScale(gpuGray, 1.1, 10, Size.Empty);
foreach (Rectangle f in learRegion)
{
//draw the face detected in the 0th (gray) channel with blue color
image.Draw(f, new Bgr(Color.Blue), 2);
using (GpuImage learImg = gpuGray.GetSubRect(f))
{
//For some reason a clone is required.
//Might be a bug of GpuCascadeClassifier in opencv
using (GpuImage clone = learImg.Clone())
{
}
}
}
}
watch.Stop();
}
}
else
{
//Read the HaarCascade objects
using (HaarCascade lear = new HaarCascade(learFileName))
{
watch = Stopwatch.StartNew();
using (Image gray = image.Convert()) //Convert it to Grayscale
{
//normalizes brightness and increases contrast of the image
gray._EqualizeHist();
//Detect the faces from the gray scale image and store the locations as rectangle
//The first dimensional is the channel
//The second dimension is the index of the rectangle in the specific channel
MCvAvgComp[] learsDetected = lear.Detect(
gray,
1.1,
10,
Emgu.CV.CvEnum.HAAR_DETECTION_TYPE.DO_CANNY_PRUNING,
new Size(20, 20));
foreach (MCvAvgComp f in learsDetected)
{
//draw the face detected in the 0th (gray) channel with blue color
image.Draw(f.rect, new Bgr(Color.Blue), 2);
//Set the region of interest on the faces
gray.ROI = f.rect;
}
}
watch.Stop();
}
}
pictureBox1.Image = image.ToBitmap();
break;
case “Hand Detection”:
if (GpuInvoke.HasCuda)
{
using (GpuCascadeClassifier hand = new GpuCascadeClassifier(handFileName))
{
watch = Stopwatch.StartNew();
using (GpuImage gpuImage = new GpuImage(image))
using (GpuImage gpuGray = gpuImage.Convert())
{
Rectangle[] handRegion = hand.DetectMultiScale(gpuGray, 1.1, 10, Size.Empty);
foreach (Rectangle f in handRegion)
{
//draw the face detected in the 0th (gray) channel with blue color
image.Draw(f, new Bgr(Color.Blue), 2);
using (GpuImage faceImg = gpuGray.GetSubRect(f))
{
//For some reason a clone is required.
//Might be a bug of GpuCascadeClassifier in opencv
using (GpuImage clone = faceImg.Clone())
{
}
}
}
}
watch.Stop();
}
}
else
{
//Read the HaarCascade objects
using (HaarCascade hand = new HaarCascade(handFileName))
{
watch = Stopwatch.StartNew();
using (Image gray = image.Convert()) //Convert it to Grayscale
{
//normalizes brightness and increases contrast of the image
gray._EqualizeHist();
//Detect the faces from the gray scale image and store the locations as rectangle
//The first dimensional is the channel
//The second dimension is the index of the rectangle in the specific channel
MCvAvgComp[] handsDetected = hand.Detect(
gray,
1.1,
10,
Emgu.CV.CvEnum.HAAR_DETECTION_TYPE.DO_CANNY_PRUNING,
new Size(20, 20));
foreach (MCvAvgComp f in handsDetected)
{
//draw the face detected in the 0th (gray) channel with blue color
image.Draw(f.rect, new Bgr(Color.Blue), 2);
//Set the region of interest on the faces
gray.ROI = f.rect;
}
}
watch.Stop();
}
}
pictureBox1.Image = image.ToBitmap();
break;
}
}
}
}
Reply
- Andol says: 12/ December/ 2011 at 8:36 PM
  @Jerry
  Hi Jerry, sorry I am quite busy these days because I have to rush up my thesis writing before Xmas holiday, so the programming work will be hung for a while until then.
  About the bug you had in your project, I guess this might be caused by some bugs in my source codes, due to some system configuration differences or programming faults. So my idea for you is, to write the code on your own, as you can still refer to the logics in my source codes. As well, please be aware of the haar training files, as these were not comprehensively compatible with the programmes.
  Good luck, cheers
  Reply
Tien Do says: 14/ December/ 2011 at 9:16 AM
Hi Andol,
I’m researching about object detection. I downloaded your xml file and used it in “performance” test of OpenCV haartraining. Unfortunately, your xml doesn’t work. I don’t understand, but I think that’s why my testing images are not good. Can you tell me about your positive folder ? If it’s secret, you should send your answer to my mail.
Thank you very much
Reply
- Andol says: 14/ December/ 2011 at 9:46 AM
  @Tien Do
  Hi Tien Do, I will try to check the haartraining xml files carefully and reply back to you later.
  Thanks for your valuable feedbacks. Would you mind sharing some of your ongoing project?
  Reply
Tien Do says: 15/ December/ 2011 at 3:13 AM
Hi Andol,
My project’s positive images just have one gesture in another angles on black background because my project use for detecting hand but not detecting gesture which is step 2. I have 2 problem :
- How can substract background in positive image. When I detected, it only detected the hand in black background
- I only used OpenCV haartraining and I want to rebuild source code in “HaarDetectingObject” function. But I really feel confused with their code ^^.
I will send you a mail with my xml file and picture of positive folder. I’m not allowed share positive images, sorry for this.
Thank you very much
Reply

detecting wire hand by hand hand in hand区别 HAND IN HAND hand in hand是什么意思？ hand in hand词曲作者 Detecting Tde drives Detecting IDE drives? detecting ide driver Detecting IDE Drivers Detecting IDE drives....... 谁有李宇春唱的《HAND IN HAND》《hand in hand》音乐元素？ Detecting IDE drives…… radia detecting and ranging是什么 Detecting IDE drires...是什么问题？ HAND IN HAND 的中文翻译歌词《 hand in hand 》伴奏带哪里有？汉城奥运会hand in hand的中文翻译哪里可以下载 hand in hand 视频? Hand in hand是谁唱的？ By the hand /with the hand? in Your Hand 是什么意思? give me a hand instant hand sanitizer是什么意思