yohanrw's profile - activity

I am asked to write a code which can detect ANY moving object using OpenCV. It will be used in in-door system. And it is real time, input will be taken from a camera, and system need to detect the moving stuff real time. According to my knowledge this is motion detection. I am not considering about unwanted movements like waving of a flag or curtain. When an important motion is detected, the system will raise an alarm.

I know this is possible with opencv, because I saw the following video

http://www.youtube.com/watch?v=lzRGp3nc8Tw

But, I do not have resources to learn how to. Official OpenCV tutorials do not contain any data about this.

Please provide me advice, links, learning resources or steps I need to follow.

@StevenPuttemans: But there are lot of android apps out there which facilitate you to watch the system online via the phone. This clearly means IP Cams has a static IP right?

Hi all,

I am about to do a cascade training in OpenCV (training for LBP Cascades) using a very large set of images. I have never done this with a large set of images, so I am interested in knowing whether it is possible to stop the training and start from where we stopped. For an example, I put the training on and it works, and turn it off at night so I can shutdown the PC, then turn it on again it in the morning and it continues the training from where it stopped. Is this is possible? It is very important for me to know this because it will decide the hardware I have to invest on and the time period. I look forward for your answers.

First of all you have to understand a bit about my background. I have never worked with IP cameras before, but I will have to access multiple IP cameras simultaneously with OpenCV very soon. Apart from that, my network knowledge is not that much good.

Now the question. I know OpenCV can access IP cameras using its URL with the help of VideoCapture. There are number of questions in StackOverflow as well. But, the IP address is something that can be changed by time to time, it is dynamic. So even in a IP cam, the IP will change by time to time. So if this is the case, does this mean every time you try to access the cameras with OpenCV, you have to find the IP address of the camera manually and change the URL in OpenCV as well? I do not need to access the cameras via the internet, accessing locally is fine, via Wifi or network cables. How can I set a fixed IP address/ URL so the OpenCV code has nothing to change every time?

PS: I navigated through lot of websites (sites from more than 5 result pages of Google search) but the answers given are either too old or not that much convincing. So I decided to walk into OpenCV forum for the answer.

@berak: First of all, I am very very sorry for the delayed reply. This forums is not giving any notifications :(. Thanks a lot for the reply, can you please submit it as the answer?

I have seen there are 2 different Haar Cascade datasets in OpenCV. For an example, take a look at haarcascade_upperbody.xml and haarcascade_mcs_upperbody.xml. what is this new mcs thing? The only difference I can monitor is that haarcascade_mcs_upperbody.xml is providing a way better results than the other one.

So, can someone please explain me the difference between these 2 types? When training my own datasets, how can I select between these 2?

@StevenPuttemans: Thanks for the reply and very sorry for the delay of my reply, I do not get email notifications! anyways, I did not understand what you mentioned. Mind providing a code sample?

wow, you are awesome! How long did it take to do this task? I mean, how many seconds it took? very sorry for the delayed reply, I never get email notifications :(

@StevenPuttemans: oh, and another thing, SimpleCV can do this, but it is in Python :(

@StevenPuttemans: Thanks for the reply. You mean use OpenCV, but use it efficiently with best ways and our own tricks?

I am wondering if there is any C++ library for reading multiple IP cameras and do the simplest motion detection. I of course know OpenCV can do this but it is kind of heavy when it comes to multiple cameras. My target is 20 IP cameras. I have seen commercial applications doing this task, so I know this is somehow possible for sure. Below are some commercial applications which can do motion detection on multiple IP cams.

1. http://www.deskshare.com/video-surveillance-software.aspx
2. http://sourceforge.net/projects/surveillance/
3. http://www.nchsoftware.com/surveillance/
4. http://download.cnet.com/Camera-Viewer-Pro/3000-2348_4-75123285.html

I am seeking for a C++ library because I need to run trigger some OpenCV code on motion detection.

Any ideas please?

Hi, My requirement has been changed from real time to detect in images. Please have a look at the updated question.

@StevenPuttermans: Hi, My requirement has been changed from real time to detect in images. Please have a look at the updated question.

Hi,

Please have a look at the below code

#include "opencv2/imgproc/imgproc.hpp"
#include "opencv2/objdetect/objdetect.hpp"
#include "opencv2/highgui/highgui.hpp"

#include <stdio.h>
#include <string.h>
#include <ctype.h>

using namespace cv;
using namespace std;

void help()
{
    printf(
        "\nDemonstrate the use of the HoG descriptor using\n"
        "  HOGDescriptor::hog.setSVMDetector(HOGDescriptor::getDefaultPeopleDetector());\n"
        "Usage:\n"
        "./peopledetect (<image_filename> | <image_list>.txt)\n\n");
}

int main()
{
    Mat img;
    char _filename[1024];


    img = imread("C:/Users/yohan/Desktop/dogwalker.jpg");



    HOGDescriptor hog;
    hog.setSVMDetector(HOGDescriptor::getDefaultPeopleDetector());
    namedWindow("people detector", 1);

    for(;;)
    {
        vector<Rect> found, found_filtered;
        double t = (double)getTickCount();
        // run the detector with default parameters. to get a higher hit-rate
        // (and more false alarms, respectively), decrease the hitThreshold and
        // groupThreshold (set groupThreshold to 0 to turn off the grouping completely).
        hog.detectMultiScale(img, found, 0, Size(8,8), Size(32,32), 1.05, 2);
        t = (double)getTickCount() - t;
        printf("tdetection time = %gms\n", t*1000./cv::getTickFrequency());
        size_t i, j;
        for( i = 0; i < found.size(); i++ )
        {
            Rect r = found[i];
            for( j = 0; j < found.size(); j++ )
                if( j != i && (r & found[j]) == r)
                    break;
            if( j == found.size() )
                found_filtered.push_back(r);
        }
        for( i = 0; i < found_filtered.size(); i++ )
        {
            Rect r = found_filtered[i];
            // the HOG detector returns slightly larger rectangles than the real objects.
            // so we slightly shrink the rectangles to get a nicer output.
            r.x += cvRound(r.width*0.1);
            r.width = cvRound(r.width*0.8);
            r.y += cvRound(r.height*0.07);
            r.height = cvRound(r.height*0.8);
            rectangle(img, r.tl(), r.br(), cv::Scalar(0,255,0), 3);
        }
        imshow("people detector", img);
        int c = waitKey(0) & 255;
            break;
    }
    return 0;
}

This is the OpenCV code for detecting humans. But I noticed that this do not detect people in most cases, for an example, please have a look at the below image.

If you run the above code on this, this will not detect the person. What is wrong here? Please help.

Hi Guys,

I have a very challenging task; detect the people in images. This is my challenge.

1. Detect people in images (but make it quick, should not take more than 5 seconds).
2. People might be walking, standing, or sitting.
3. The Race of the people should not matter.
4. The background environment should not matter. (ex: Person might be sitting next to Home, Person might be standing infront of the car, people might be walking in beach)

How can I achieve this task? I can only think of Cascades. Any ideas? How good the Cascade training should be? any other ideas?

Hi,

Anyone know a good dataset of images which we can use for people detection? All the databases I found contains only the positive images, no negative images. And in some cases, it contains images of people standing only. Please help.

Thank you. I appreciate your help.

Please have a look at the below code

#include <opencv2/highgui/highgui.hpp>
#include <iostream>

using namespace cv;
using namespace std;

int main()

{
    cv::VideoCapture cam1,cam2;
    cam1.open(0);
    //cam2.open(0);

    Mat im,im2;

    cam1>>im;
    cam1>>im2;

    while(true)
    {
        cam1>>im;

        for(int i=0;i<15000;i++)
        {


        }
        cam1>>im2;

        Mat im3 = im2-im;

        imshow("video",im3);

        if(waitKey(30)>=0)
        {
            break;
        }

    }
    waitKey(0); 
}

I am trying to identify the difference (in other terms, motion) by subtracting the images. However what I get is a 100% blank screen. If I use 2 VideoCapture instances capture frames and load them to im and im2, then it works. But I must not use 2 VideoCapture instances, I must only use 1. what have I done wrong here?

I am seeking for a library which allows me to detect "people" in real time using the camera. Multi camera support is more preferred. I looked into lot of libraries including OpenCV but the case is either they have patented algorithms, or has license like GPL which restricts the commercial use. In my case, I need to make a commercial product, source code is closed, and will be distributed commercially.

I prefer to have a free library. Paid libraries also can be considered. Preferred languages are C++, C#.Net or Java.

Please share your knowledge with me.

Hi,

I have 1 question here.

1 - I am wondering whether we can use OpenCV commercially. I am creating a program which will be distributed commercially, and it will be based on OpenCV. I didn't understand anything by reading the license, because it is all about pure law.

So my question is, Can we use OpenCV in a commercial application? Is there any restrictions? I appreciate to have the answer in pure English, so anyone who dnt understand law documents will understand it.

Please help.

@berak: You mean the softcascades are not supported any more?