hello
I installed code block and I copy and paste the
code

include "opencv2/opencv.hpp"

using namespace std;
using namespace cv;

int main( int argc, char* argv[] )
{
// Load in test image
Mat image = imread("/home/koj/123jpeg", IMREAD_GRAYSCALE);
imshow("original", image); waitKey(0);

return 0;

}

and went to" build option" as recommended by steven and "linker setting" ->"other linker options" ->I typed in the "pkg-config opencv --libs"
and started build and run
the result was hello world inthe console screen which I could close by entering some key.

1.is that right???
2.where can I use the command $ g++ test.cpp -o test `pkg-config opencv --libs'?

it is very frustrating .
3.what project should I make to begin with? opencv project or console application project.
4.and why does the project file name becomes cbp instead of cpp?

Hi,

I'm reading the book and trying to implement the code in chapter 5 to plot the PRC curve for a classifier correctly.

I don't understand why, but after doing the annotation and generate the output of the classifier detection I am obtaining values very similar to this Link

I cannot understand what is wrong, can you help me with this?

If needed I can provide the used annotation and output txt files.

Regards

I installed g++ and when i go into the code block and open a console application project.
ANd I seleted C++ as my language and then the next page shows gnu GCC compiler and all the other things except g++.
I heard that I should have g++ compiler to follow this book but why the code block does not detect installed g++?
I searched google for hours but everybody just says about installing build essential which i already did two days ago.
Still struggling for installation
give me some help.

I am already using opencv on my windows 64bit PC. I had installed it using CMake and Visual Studio. Now, I want to add the source code this repo to my already existing OpenCV directory. How should I proceed ?

Hi guys! As the title says, this is about Chapter 5 of the book, so this is mainly for you @StevenPuttemans

1. Reporting a possible trivial errata. The cascade classification process in detail > Step2 > First paragraph: [...] especially when knowing that a model of 24x24 pixels can yield more than 16,000 features -> Shouldn't it be 160,000 features? Not the most important errata at all, but just letting you know.

2. A question about the explanation of the maxFalseAlarmRate param. I'm not sure if this is an errata, a confusing explanation, or just that I'm plain wrong. I'll start by copying the book explanation:

- maxFalseAlarmRate: This is the threshold that defines how much of your negative samples need to be classified as negatives before the boosting process should stop adding weak classifiers to the current stage. The default value is 0.5 and ensures that a stage of weak classifier will only do slightly better than random guessing on the negative samples. Increasing this value too much could lead to a single stage that already filters out most of your given windows, resulting in a very slow model at detection time due to the vast amount of features that need to be validated for each window. This will simply remove the large advantage of the concept of early window rejection.

So, what's confusing me is the second half the of paragraph. You say increasing the param value results in more filtering and more number of features to validate per stage. As I understand it, increasing the maxFalseAlarmRate implies letting the stage to misclassify more negative samples as positive ones. Therefore, less negative samples are rejected in that stage, and a higher number of accepted windows passes to the next stage. Also, I think (though I'm not too sure here) that the number of weak classifiers/features per stage is lower as the maxFalseAlarmRate is higher. In fact, as you say later on, This process continues until the false acceptance ratio on the negative samples is lower than the maximum false alarm rate set. Therefore, with a higher maxFalseAlarmRate, the chances to get a lower acceptance ratio increases, leading to stages with less features.

To sum up, did you really want to say increasing this value instead of decreasing this value, or did I simply get it wrong?

3. An additional doubt (I can post it in the OpenCV forum if you prefer). Based on your experience, is there any con (at detection time) of using more number of stages, but with less restrictive conditions? e.g. using 20 stages with minHit 0.990 instead of using 10 stages with minHit 0.995 (supposing you have enough data to tackle training in both cases, and that the final minHit is the same -haven't done math-)

4. Good job on the chapter ;)

Hi,

I am running through chapter 2 with my Canon PowerShot G9, which has the following configurations for exposurecompensation.

/main/capturesettings/exposurecompensation
Label: Exposure Compensation
Type: RADIO
Current: +1 2/3
Choice: 0 +2
Choice: 1 +1 2/3
Choice: 2 +1 1/3
Choice: 3 +1
Choice: 4 +2/3
Choice: 5 +1/3
Choice: 6 0
Choice: 7 -1/3
Choice: 8 -2/3
Choice: 9 -1
Choice: 10 -1 1/3
Choice: 11 -1 2/3
Choice: 12 -2

So I need to mofify capture_exposure_bracket.sh, perhaps starting from somewhere below 6, and to going smaller.

Could you suggest where should I start, and end by default? Like, 9, 6, 3?

Thanks!

when running "fingerprint_process", I found orb_descriptor that's under "feature2d.hpp",
seems "xfeatures2d.hpp" does not use? (my opencv version is 3.1.0)
And "orb_descriptor->compute(input_thinned, keypoints, descriptors);"

this line will throw a exception :

C:\buildslave64\win64_amdocl\master_PackSlave-win64-vc14-shared\opencv\modules\features2d\src\orb.cpp:997: error: (-215) level >= 0 in function cv::ORB_Impl::detectAndCompute

I don't know what they mean...can't solve it.
Does I use the wrong library? This version's method has some changes?
Would anyone help me~Please!

When testing method "meshwarp" from "python/utilites.py", you use some file from the folder "/work/blueprints/c7/data/mesh2.png". Unfortunately, it is not attached to this chapter as data. Therefore user can not test this method with your test data.

https://github.com/OpenCVBlueprints/OpenCVBlueprints/blob/master/chapter_7/python/utilities.py#L129

I tried using sudo too. The program works, but it does not write to the annotations file. I noticed in another issue that this program should no longer be used. Would be great if the updatable book said that!

If someone knows how to get it working, please let me know. I see there's an OpenCV provided tool, but haven't found the params for that yet.

What's happening

Preprocessing happens on undefined Mat objects. It will not make a difference on the output result.

What should be happening

Preprocessing should be happening in the for loop to be effective.

Environment

• Arch Linux x64
• CLion 2018.6
• OpenCV 3.4.3

Link

https://github.com/OpenCVBlueprints/OpenCVBlueprints/blob/master/chapter_5/source_code/rotation_invariant_detection/rotation_invariant_object_detection.cpp#L82-L95

Hi,

I'm running the python code given in chapter 7 after importing the video and gyroscope files on a mac. The application crashes in the meshwarp function with the following stack trace.

Traceback (most recent call last):
  File "/Users/jeetdholakia/PycharmProjects/HyperlapseTest/stable.py", line 601, in <module>
    stabilize_video(mp4path, csvpath)
  File "/Users/jeetdholakia/PycharmProjects/HyperlapseTest/stable.py", line 577, in stabilize_video
    prev, focal_length, gyro_delay, gyro_drift);
  File "/Users/jeetdholakia/PycharmProjects/HyperlapseTest/stable.py", line 265, in accumulateRotation
    o = utilities.meshwarp(src, pts_transformed)
  File "/Users/jeetdholakia/PycharmProjects/HyperlapseTest/utilities.py", line 120, in meshwarp
    mapx = np.append([], [ar[:,0] for ar in g_out]).reshape(mapsize).astype('float32')
ValueError: cannot reshape array of size 2332800 into shape (1080,1080,1)

The line which produces the error is:

utilities.py -> def meshwarp(src, distorted_grid)
mapx = np.append([], [ar[:,0] for ar in g_out]).reshape(mapsize).astype('float32')

P.S. - There are small errors in other classes as well, which required minor changes in code to get it up and running, they include :

1. Removing calibration_worker import (module not found error!) and importing from calibration import GyroscopeDataFile to make the gyroscope data work.
2. The gyroscope file generated does not have the extension .gyro.csv, its just XYZgyro.csv, where as python code looks for the aforementioned extension.
3. pts_transformed variable in def accumulateRotation, is not initialized anywhere, hence i initialized it as an empty array pts_transformed = [] in the beginning of that function.
4. In the same function, the supplied code output = cv2.perspectiveTransform(np.array([[pixel_x, pixel_y]], transform) throws up numpy error data type could not be understood. Hence, I changed it to output = cv2.perspectiveTransform(np.array([[pixel_x, pixel_y]], dtype="float32"),transform), and the code worked.

Please let me know how to fix the aforementioned issue, and if I have made any mistakes while correcting the code which threw some compile time and run time errors.

Thanks,
Jeet

Following the instructions, I

1. created a directory in "C:" called "CVmodules" where I put the zips downloaded from the Itseez link(opencv-master and opencv_contrib-master).
2. In the Cmake GUI selected the opencv source code folder (C:\CVmodules\opencv) and the folder where binaries will be built (C:\CVmodules\contribBuild)
3. pressed the configure button.
4. browsed the parameters in the interface and look for the form called OPENCV_EXTRA_MODULES_PATH
5. completed this OPENCV_EXTRA_MODULES_PATH by the proper pathname to the <opencv_contrib>/modules value (C:\CVmodules\opencv_contrib\modules).
6. then pressed the configure button

However, the picture below is all I see when research tessearct. There is no way to enable tesseract.
I also realized that there seem not to be a tesseract library, in the tesseract that I installed,to point it to. I downloaded it from
https://code.google.com/p/tesseract-ocr/downloads/detail?name=tesseract-ocr-setup-3.02.02.exe&can=2&q=

What am I doing wrong? Did download the wrong version of the tesseract. This is for visual studio 2013 in windows 8. I already have opencv 3 installed using the prebuilt libraries method. Should uninstall and build it instead?

Hi, thanks for those detailed instructions.

My env is Ubuntu 14.04.3.

apt-get install build-essentials

This is supposed to be:

apt-get install build-essential

And this is:

apt-get install cmake cmake-qt-gui git pkg-config libgtk2.0-dev libavcodec-dev libformat-dev libswscale-dev

supposed to be:

apt-get install cmake cmake-qt-gui git pkg-config libgtk2.0-dev libavcodec-dev libavformat-dev libswscale-dev

One more. libeigan-dev not found.

apt-get install libeigen-dev libeigen2-dev libeigen3-dev

No alternative I came up with.

apt-get install libeigen2-dev libeigen3-dev

Also, I suggest to add sudo for Ubuntu installation commands.

I read the step by step guide and I am stuck at the directory explanation.
my root folder is " /home/koj"
when I made a directory named "opencv" in that root directory and inside that "build" and inside that "opencv_contrib" as explained in the guide.
and then if I try
git clone https://github.com/Itseez/opencv.git opencv
then there comes an error that there already exist opencv directory and not empty
So I executed
git clone https://github.com/Itseez/opencv.git opencv
in the root directory (" /home/koj") without creating (a directory named "opencv" and inside that "build" and inside that "opencv_contrib") that directory chain and
after downloading the file in root directory I could find that there is an "opencv" directory made by downloading itself.
So I add "build" directory and then at that build directory I executed
git clone https://github.com/Itseez/opencv_contrib.git opencv_contrib
then there was automatically made directory in the "build" directory named "opencv_contrib".

So after that I went to build directory and executed "cmake-gui .."
but the return was cmake-gui:cannot connect to x server

And I cannot find any answer at googling.
Please help me to install opencv.
I am at chapter 1 and cannot progress anymore.

hello
I cannot understand the last part of the installation guide.
I opened gedit and pasted the test code with the image path replaced with "/media/koj/Ubuntu-Server 16.04.1 LTS amd64"
and I activated external tools at gedit so there appeared "build" at the "external tools"
so when i pressed build at the gedt
Running tool: Build

No Makefile found!

Done
can you explain why this happen? and what does it mean in the below which says "configure your editor to include the build and opencv repositories"

Make sure that you replace the image path with an actual image path that exists! Then, configure your editor to include the build and installed OpenCV repositories. Under Ubuntu, you can do that by pointing your linker to

pkg-config opencv --libs
which will auto grab all the dependencies. This can be done inside your IDE on the linker settings tab or by executing the following command

$ g++ test.cpp -o test pkg-config opencv --libs
using the command line. Since all headers are installed into /usr/local/lib/, you do not need to explicitly tell that to the system, since Ubuntu can handle that itself.

I have added
1> "ndk.dir" variable in local.properties file.
2 > Change the absolute path to OpenCV SDK

In JNI we have com_example_panorama_NativePanorama.cpp file.
In that we have
#include opencv2\opencv.hpp"
#include opencv2\stitching.hpp"

but in our jni folder we dont have opencv.
How to solve this issue?

hello
i bought playstation eye and inserted to my notebook which has vmware pro12 and ubuntu 14.04 installed.
I cannot find how can i install my playstation eye driver in this book to my ubuntu so i just connected to my usb slot and i cannot figure out it is working or not.
and I copied the cpp file of the blinking eye source code and then I made a console project at codeblocks.
and then build and run .
First time it made an error so I add to linker setting the pkg-config opencv --libs
and then it made no error

when i ran it first time it made a blank x terminal with no message i waited but nothing appeared.
So I quit the terminal by clicking x button in the terminal window and when I go back to the project folder there was a slowmo video file.but when i played it says it is empty .
So I thought there is a problem with the playstation eye installation.
So I searched google and find out following site
http://oftw.wikidot.com/install-ps3-eye-camera-in-ubuntu
but it did not worked at second the line
"sudo patch -p1 < ps3eyeMT-2.6.31-10-generic.patch
only patch the driver in drivers/media/video/gspca/ov534.c"
I could not understand the meaning of "only patch the driver ..."
So it made an error at console window and i gave up following that method.
So I came back to my codeblock and run again this time the following attached screen appeared at the window.
So I changed the frame rate in the source code to 120 but this time again the same window appeared.
how can I install the playstation eye driver to my computer and why does this slowmo video is empty?
how do i know my playstation eye is working in ubuntu.?
what is the playstation eye dial at the lens functions is it a power button?

Hello, I am trying to follow along with chapter 5 and am having problems with the object annotation code. I am using Ubuntu 14.04.05 and have installed opencv 3.1. I cloned the entire OpenCVBlueprints rep. and once inside the object annotation dir I compiled the code by doing

cmake .
make

Then I made sure that I had some images handy in the ~/Pictures dir.

When I execute the command

./object_annotation -images ~/Pictures/ -annotations ~/Pictures/annotation.txt

the terminal simply returns to a new prompt with out providing any errors or feedback.

However if I run the code provided by opencv

./opencv_annotation -images ~/Pictures/ -annotations ~/Pictures/annotation.txt

everything works as expected

Any Ideas?

Thanks,

Ryan

Hi,

Under Ubuntu, you can do that by pointing your linker to

pkg-config opencv --libs

which will auto grab all the dependencies.

Is this what you meant?

$ g++ test.cpp -o test `pkg-config opencv --libs`

Then, writing so is more helpful, though it might be obvious.

Good day,

I'm getting the following error in Chapter 7: UnboundLocalError: local variable 'rot_end' referenced before assignment

This seems to happen in stable.py in the following function:

def getAccumulatedRotation(w, h, theta_x, theta_y, theta_z, timestamps, prev, current, f, gyro_delay=None, gyro_drift=None, shutter_duration=None, doSub=False):

    if not gyro_delay:
        gyro_delay = 0

    if not gyro_drift:
        gyro_drift = (0, 0, 0)

    if not shutter_duration:
        shutter_duration = 0

    x = np.array([[1, 0, -w/2],
                      [0, 1, -h/2],
                      [0, 0, 0],
                      [0, 0, 1]])
    A1 = np.asmatrix(x)
    transform = A1.copy()

    prev = prev + gyro_delay
    current = current + gyro_delay

    if prev in timestamps and current in timestamps:
        start_timestamp = prev
        end_timestamp = current
    else:
        (rot_start, start_timestamp, t_next) = fetch_closest_trio(theta_x, theta_y, theta_z, timestamps, prev)
        (rot_end, end_timestamp, t_next) = fetch_closest_trio(theta_x, theta_y, theta_z, timestamps, current)

    gyro_drifted = (float(rot_end[0] + gyro_drift[0]),
                    float(rot_end[1] + gyro_drift[1]),
                    float(rot_end[2] + gyro_drift[2]))
    if doSub:
        gyro_drifted = (gyro_drifted[0] - rot_start[0],
                        gyro_drifted[1] - rot_start[1],
                        gyro_drifted[2] - rot_start[2])
    R = getRodrigues(gyro_drifted[1], -gyro_drifted[0], -gyro_drifted[2])

    x = np.array([[1.0, 0, 0, 0],
                     [0, 1.0, 0, 0],
                     [0, 0, 1.0, f],
                     [0, 0, 0, 1.0]])
    T = np.asmatrix(x)
    x = np.array([[f, 0, w/2, 0],
                  [0, f, h/2, 0],
                  [0, 0, 1, 0]])
    transform = R*(T*transform)

    A2 = np.asmatrix(x)

    transform = A2 * transform

    return transform

Any ideas why rot_end would be empty or am I missing something related to Python3?

please try the unblinking eye code with ubuntu 16.04 again.
changin camera index to 0 and 1 gives the same result.
and now cheese program is working with photo, video and burst mode.

and another question is at the asus xtion pro running page.
I could not understand the following command how should i type this command is which folder what does the backslash means at the end of each line ?

hello again
I installed asus xtion pro .
one thing i noticed was the command line of the book regarding usb3.0 and usb2.0 problem did not work.
i tried wine cmd with running the usbupdate.bat file and it worked.
And another issue was installing openni2 the installation guide maybe outdated that openjdk-6-jdk nolonger exist and openjdk-8-jre was what i have done.
and when i added all the library files in the linker setting I added 7 which was in the folder which was pointed by the author in the book.
and included the search directory two of them openni2 and opencv .
and ran the infravision and it run good
but when i want to exit the program so i clicked the red button on the infrared camera screen.
but it reappeared soon and the console was saying some error message about the usb priority problem.
I did it both on the vmshare and ubutu alone computer but both had exactly the same problem.
when i turned off the console screen the infrared filming screen and console was gone but the red light of the xtion pro which i think is indicating the ir light is emitting is stlll on.
So i had to pull off the usb cord from my computer to trun off those red light of asus xtion pro.
do you know why this happens?
i attached some photo

Hello,

I am trying to get the fingerprint functionality up and running (OpenCVBlueprints/chapter_6/source_code/fingerprint/fingerprint_process/fingerprint_process.cpp). The file shows OpenCV2 is required.

I tried using OpenCV 2.4, however it did not work with the fingerprint algorithm. Could someone point me to the correct OpenCV version i should use?

Thanks,
Mike

The book text in page 190 calls for a -images switch, but the code is looking for a -output switch

Book:

First, parse the content of your positive image folder to a le (by using the supplied folder_listing software inside the object annotation folder), and then follow this by executing the annotation command:
./folder_listing –folder <folder> -images <images.txt>

Code:

Line 38 in folder_listing.cpp

        }else if( !strcmp( argv[i], "-output" ) )
        {
            output = argv[++i];
        }

Good day,

I'm currently working on Chapter 7. I've created an Android application to capture video and write to the gyro file.

When I try to use the captured video and gyro file with stable.py I get the following error:

  File "stable.py", line 572, in <module>
    stabilize_video(mp4path, csvpath)
  File "stable.py", line 548, in stabilize_video
    rot = accumulateRotation(frame, delta_theta[0], delta_theta[1], delta_theta[2], timestamps, previous_timestamp, prev, focal_length, gyro_delay, gyro_drift)
  File "stable.py", line 237, in accumulateRotation
    output = cv2.perspectiveTransform(np.array([[pixel_x, pixel_y]], dtype="float32"), transform)
cv2.error: /tmp/opencv-20170224-1869-10nlf6f/opencv-2.4.13.2/modules/core/src/matmul.cpp:1943: error: (-215) scn + 1 == m.cols && (depth == CV_32F || depth == CV_64F) in function perspectiveTransform

Any idea as to what the problem could be?

Hi,

After terminating set_motion_trap.py by interrupting with Ctrl+C, my camera stays busy, and won't respond with the following error.

$ gphoto2 --list-all-config

*** Error ***              
An error occurred in the io-library ('Could not claim the USB device'): Could not claim interface 0 (Device or resource busy). Make sure no other program (gvfs-gphoto2-volume-monitor) or kernel module (such as sdc2xx, stv680, spca50x) is using the device and you have read/write access to the device.
*** Error (-53: 'Could not claim the USB device') ***       

For debugging messages, please use the --debug option.
Debugging messages may help finding a solution to your problem.
If you intend to send any error or debug messages to the gphoto
developer mailing list <[email protected]>, please run
gphoto2 as follows:

    env LANG=C gphoto2 --debug --debug-logfile=my-logfile.txt --list-all-config

Please make sure there is sufficient quoting around the arguments.

After manually plug it out and in, it starts responding again.

Is there anyway to cleanly terminate the program? Or anything wrong?

Thanks.