zshn25's profile - activity

Having the same issue. Did you find any solution for this?

I suggest you first try your app on PC, rather than jumping to Android.

The problems you face are common ones if you are using any of the OpenCV Face Recognition algorithms (LBPHFaces, FisherFaces, EigenFaces). Try using all of these. In your PersonRecognizer.java file, line 16, change createLBPHFaceRecognizer to createFisherFaceRecognizer or createEigenFaceRecognizer and see which works best for you.

As per the face inside face problem, once you detect face, you can restrict the blobs according to its size(length, width, area). As for the beard part, try using other cascade classifiers. If not, create your own.

Research on how you can improve the available techniques or implement another Face recognition algorithm. (Read related papers)

Turns out, the number plate is not actually white. I tuned the HSV parameters a little bit and it worked.

Also, this is not a reliable method for detecting number plates. Since, it changes from image to image.

The aspect ratio of cars and moterbikes may be relatively less. But, cars will have more width (and length) You can check the length and width of each blob and set a threshold to classify the vehicles.

Also, you can check the blob properties in an ROI (the junction) to make it more effective. The blob area and aspect ratio wouldn't vary that much within the ROI

Thanks, that worked :)

But why is it not building. May I ask? Did I do anything wrong?

I downloaded and built both Tesseract as well as Protocol buffer from source. But CMake gives me error. Tesseract: NO Could NOT find PROTOBUF (missing: PROTOBUF_LIBRARY) Build libprotobuf from sources: libprotobuf not found into system Tesseract: NO

If you just want to run it, then this code will directly run with OpenCV 3.0.0. But using the depreciated C functions is not recommended. Also, I think the problem is in including the libraries. How are you compiling this? If you are using Visual Studio in Windows, then follow this tutorial to compile.

One method for avoiding false positives might be to check for the size of the detected blob, or to check its aspect ratio. Faces must have a particular aspect ratio and thus eliminates false positives

Maybe it is am ffmpeg problem, try copying the ffmpeg ddl file from opencv built folder to your application folder

In the video path, give '/' instead of '\'

Please stop using the depriciated C code. For example use Mat instead of IplImage, etc.That should do it

And, yes, you can add long code. Edit your question and add the code to it in the end

Then maybe the problem is something else. Help me help you by posting your code

#include <opencv2/opencv.hpp>

should do it

I think your rtsp link is broken. I tried the same with OpenCV 3.0.0 and its working perfectly fine!

I am trying to detect the number plate in an image. Usually licence plates in my country are black texts with white or yellow background. So, I detect these colors from an image. But my algo gives me everything except the licence plate. The code I used is

Mat img_hsv;
cvtColor(input_image, img_hsv, CV_BGR2HSV);

Mat img_yellow;
int lowerHue = 22;  //For yellow
int upperHue = 38;
inRange(img_hsv, Scalar(lowerHue, 100, 100), Scalar(upperHue, 255, 255), img_yellow);

Mat img_white;
inRange(img_hsv, Scalar(0, 0, 200), Scalar(255, 55, 255), img_white);

Mat img_black;
inRange(img_hsv, Scalar(0, 0, 0), Scalar(255, 55, 50), img_black);

Mat mask = img_black + img_white + img_yellow;
Mat img_result;
bitwise_and(input_image, input_image, img_result, mask);

and the result is

Could you please give us more info? Is the rtsp link working? You can check that through vlc player. What error are you getting?

I guess updating your graphics card drivers might help.

I'm sorry, it worked. Thanks!!

Let me tell you something about headers, When you #include < >, you are including standard library header files. For example #include <sstream>. It includes the system header files. When you #include " ", you are including user-defined header files. For example #include "ros/ros.h". These are the libraries you downloaded.

The python function getRotationMatrix2D as follows

center = tuple(np.array(image.shape)/2)
rot_mat = cv2.getRotationMatrix2D(center,angle,1.0)
result = cv2.warpAffine(image, rot_mat, image.shape,flags=cv2.INTER_LINEAR)

gives this error

Traceback (most recent call last):
File "C:\Users\zeeshan khan\Desktop\opencv\Code\Python\rotate.py", line 13, in <module>
rot_mat = cv2.getRotationMatrix2D(center,angle,1.0)
TypeError: function takes exactly 2 arguments (3 given)

How about strings.h? (the plural of string)

I tried running the super resolution OpenCV sample. I built OpenCV 3.0.0 with CUDA on my 64-bit machine. The function cv::superres::createFrameSource_Video_CUDA gives the following error

OpenCV Error: The function/feature is not implemented (The called functionality is disabled for current build or platform) in throw_no_cuda, file C:\opencv\source\modules\core\include\opencv2/core/private.cuda.hpp, line 101

So, it uses createFrameSource_Video function to get the input. Also, other CUDA functions work fine. Just the above one gives an error.

It works fine with the farneback Optical flow algoritm but does't work with others. For example, the function createOptFlow_DualTVL1_CUDA gives the following error

OpenCV Error: Assertion failed (allocSize == allocations.back()) in `anonymous-namespace'::MemoryStack::returnMemory, file C:\opencv\source\modules\core\src\cuda_stream.cpp, line 99

Use #include <string> or #include <string.h>.

The solution is that you should use <..> instead of '..'.

Hope it helps

You can just use

cv::VideoCapture cap(0);
cv::Mat frame;
while(cap.read (frame)){
  cv::cvtColor(frame, frame, BGR2RGB);
  cv::imshow("RGB", frame);
  cv::waitKey(1);
}

Try replacing \ with / in your path to image. Also, try copying the image in your program folder and just give

im = cv2.imread("image.jpg")

cv2.imshow("", im)

cv2.waitKey(0)

Try updating your Nvidia device drivers. Worked for me

For the first two errors, its because of the CV version 3. Try this

Ptr<descriptorextractor> extractor = SURF::create();

Ptr<featuredetector> detector = SURF::create();