JohannesZ's profile - activity

Lorena, you should read the article in depth - of course the non-maxima supression is expressed in detail, also with code provided. Nevertheless I think the article can give some hints how to improve the results.

Hi, I think this post may help you. It discusses the parameters in detail:

http://www.pyimagesearch.com/2015/11/...

Sure, it's openCV with Python, but the used parameters should be the same... Johannes

Hi there,

dealing with the stitcher module in openCV 3.0 (newest checkout from GitHub), Win7 32bit and VS 2013, and searching this forum I did not find a reasonable answer yet. A straighforward implementation for my problem:

cv::Mat imgLeft = cv::imread("panoLeftS.jpg");
cv::Mat imgRight = cv::imread("panoRightS.jpg");

std::vector<cv::Mat> images;
images.push_back(imgLeft);
images.push_back(imgRight);

cv::Mat pano;
cv::Stitcher stitcher = cv::Stitcher::createDefault(false);

stitcher.setRegistrationResol(-1); // 0.6
stitcher.setSeamEstimationResol(-1);   // 0.1
stitcher.setCompositingResol(-1);   //1
stitcher.setPanoConfidenceThresh(-1);   //1
stitcher.setWaveCorrection(true);
stitcher.setWaveCorrectKind(cv::detail::WAVE_CORRECT_HORIZ);

cv::Stitcher::Status status = stitcher.stitch(images, pano);

if (status != cv::Stitcher::OK)
{
    std::cout << "Can't stitch images, error code = " << status << std::endl;
    return -1;
}

In debug mode, everything is working fine. Running in release mode I am always getting a segfault in this line (ocl.cpp) without any detailed information:

bool allocate(UMatData* u, int accessFlags, UMatUsageFlags usageFlags) const
    {
        if(!u)
            return false;

        UMatDataAutoLock lock(u);

        //here is the crash (!)
        if(u->handle == 0)
        {
            CV_Assert(u->origdata != 0);
            Context& ctx = Context::getDefault();
            int createFlags = 0, flags0 = 0;
            getBestFlags(ctx, accessFlags, usageFlags, createFlags, flags0);

What I am doing wrong? On my two machines, I set the OPENCV_OPENCL_DEVICE = Intel(R) OpenCL:CPU:0 and installed the Intel openCL driver set.

Do you have any hints what could be the problem??

Thanks for your help! Johannes

Would be nice: PTAM, DTAM, a better FLANN documentation :-)

also: http://answers.opencv.org/question/59...

and http://answers.opencv.org/question/62...

after digging deeper into the headers, I found the following:

type deviceType

• 1 default
• 2 CPU
• 4 GPU
• 8 Accelerator - whatever
• 65540 dGPU - deticated GPU
• 131076 iGPU - integrated GPU
• 1 All

Will post my results, thanks in advance!

Yes this could be a problem with the Core 2 Duo. Anyway, what do you think about my two points mentioned?

I am using a Core 2 Duo Processor with compute capability 1.2.

On another system with a new Xeon processor and and integrated Intel GPU I get the following:

• when I am iterating trough all devices with context.create(cv::ocl::Device::TYPE_ALL)), only the CPU is detected. Splitting the search in CPU and GPU from the code above I can see both devices.

• to what refers the attribute "type" of a found device? i though that the original mapping would be:

  0 -> default; 1 -> CPU; 2 -> GPU; 3 -> Accelerator; 4 -> dGPU; 5 -> iGPU; 6->ALL.

  But from this point of view, my XEON CPU is specified as type = 2, means a GPU. What is wrong here?

Hi, it is a GeForce GTX 550Ti, not the newest model, I know. OpenCL compute capability is 1.1. This could be too low for openCV, right? Iterating trough the platform info (please see my post before) I was able to see the accelerator, with the context mentioned above not. This is somewhat strange...

Hm, the context GPU works well both in debug and release. However, the context CPU throws an exception in debug mode:

"C:\builds\master_PackSlave-win32-vc12-shared\opencv\modules\core\src\ocl.cpp:2530: error: (-215) clGetDeviceIDs_pfn( pl, dtype, 0, 0, &nd0 ) == 0 in function cv::ocl::Context::Impl::Impl

in release mode, the CPU context works, but no devices are found... this is strange.

Hi unbehagelig,

thanks for your answer and your code, I will test it soon! Did you also set the environment variable for openCL OPENCV_OPENCL_DEVICE. if yes, what kind of value did you set?

Thanks for your help, JoZ

Your question has been discussed here for several times - please use the search function of this forum.

Please use the search functionality of this forum. This optic ist asked quiet often.

First, the number of points depends always on your texture. SIFT/SURF extract their features in corner-like image structures. So no corners, no/less feature points of poor quality.

In general, there is no hard border for match / non match. You may set a specific threshold but this wilk not give you robust results. A solution is to work with probabilities. Imagine a histogramm which tells you "model A gets 20 matches", "model B gets 2 matches" and "model C gets 3 matches". In this case, your distribution will tell you what will be a good match. Also try to look for terms like "bag of words classifier", "k-means" etc. OpenCV will help you with this.

what about timing? Did the images "change" their content between the shots? Moving clouds could be a good reason - the other parts were static.

At least a model for bag of words is available.

What do you recommend as a OpenCL Runtime for Intel Processors (Core Processors)?

For example, is the "Intel Integrated Native Developer Experience Starter Edition" suitable? Or ist there anoher software package from Intel recommended?

yes sure. But in case of the environment variable, the methd useOpenCl always returns false, no matter of setuseOpenCL(true/false)

Hm, this is strange: without the environment variable you mentioned, cv::ocl::useOpenCL() always return true.

When i set the environment variable to:

Intel(R) OpenCL:CPU:0

There are no error messages (I recently got an error message writing 1:CPU:0) So these parameters should be fine. But in this case, useOpenCL() always returns false.

This is somewhat strange!

Thanks for your tip! I think it's too bad that you can't set the device in the code directly. I made some tests (see below) but the cast to the void* pointer is not working - no idea why.

The conversion is not working - both ways not:

//OpenCL:: Platform
const cv::ocl::PlatformInfo* platform = &platforms[0];

//Access known device
cv::ocl::Device current_device;
platform->getDevice(current_device, 0);

//set Device - NOT WORKING
cv::ocl::Device::set(&current_device);

or

//set Device - NOT WORKING

void *p;
p = (void*)& current_device;
cv::ocl::Device::set(p);

Hi theodore!

Thanks for your answer. As I understand correctly the current API von openCV 3.0 beta has some changes in the API. Actually, I am retieviing the platform info and device info using

//OpenCL: Platform Info
std::vector<cv::ocl::PlatformInfo> platforms;
cv::ocl::getPlatfomsInfo(platforms);

//OpenCL Platforms
for (size_t i = 0; i < platforms.size(); i++)
{
    //Access to Platform
    const cv::ocl::PlatformInfo* platform = &platforms[i];

//Platform Name
std::cout << "Platform Name: " << platform->name().c_str() << "\n";

//Access Device within Platform
cv::ocl::Device current_device;
for (int j = 0; j < platform->deviceNumber(); j++)
{
    //Access Device
    platform->getDevice(current_device, j);

    //Device Type
    int deviceType = current_device.type();

In the api, there is only a method like

void cv::ocl::Device::set   (void * d)

from: http://docs.opencv.org/master/d7/d9f/...

My question is, how can I use it? I really don't understand the parameter or the cast to void*. Any help would be appreciated!

Hi,

in OpenCL, I have studied the ocl_test.cpp. Retrieving information about my system (Windows 7 SP1, 32bit) concerning available CPUs, GPUs and so on works fine. Studying the documentation, I have not found a way to set the device to do the actual computations.

In an older version of openCV, I saw a method like "setDevice", which is not available in the 3.0 beta. So, how can I enable a specific device for further computations? I only found a method called "set" which accepts a void-pointer. Since my GPU (device 0) is not the newest model I would like to enable my multicore-CPU (following device 1 and so on).

Thanks for your tips! Johannes

Hi, there was a failure in my path variable, I linked my executable against an older version of openCV. It is working now, thanks a lot!

Hmm, maybe a stupid question but what kind of version is this exactly?? 2.4.X oder a 3.0 beta?

openCV at GitHub

Hi, thanks for your answer. The problem is that I definitely don't want to use CUDA, so I set all CUDA-related flags to OFF/NO. I really don't understad why I get some kind of "cuda error message" mentioned above.

Hi there,

again a new post about compilation from openCV 3.0 master branch and tbb. Here is what I have done so far:

• machine: win7, 32bits, visual studio 2013 (vc12)
• set all path variables to openCV (code from master branch) and TBB (precompiled-win-zip)
• ran cMake and activated with_tbb.
• After configure, set the path to the relevant files (include-dirs, lib-dir, stddef-path)

• my openCV folder is located at C:\Code\openCV3, my TBB folder C:\Code\tbb - is this a problem or must the tbb folder be included directly in the openCV folder?

• added the tbb include and lib files to a property sheet and build openCV as debug (later as release). No errors, everything worked fine.

• hit "install" including and linking to all files
• in my test application compilation works, but at execution time, I am getting this failure:

"Failure in a procedure entry point" "?getHostMemRef@_OutputArray@cv@@UBEAAVHostMem@cuda@2@xz was not found in the dll opencv_world300d.dll"

I definitely deactivated WITH_CUDA and WITH_CUFFT in Cmake. So, why do I get this failure and what does "cuda" in this endless string mean? I have no CUDA graphics card and therefore deactivated these options, so I don't understand this problem.

Do you have a tip for me? Thank you so much, Johannes

Short answer: http://szeliski.org/Book/

Long answer: you should read this book. Besides, look for Zisserman & Hartley, David Forsyth.

9.7 Kilobytes?! This is too less for a raw image file, thats for sure. Even for JPEG to less... try to load the jpeg file with the code above and see what's happening. Maybe its some kind of a thumbnail image ;-)