I double checked and it is perfect now! thankyou so much. I am doing this for a project at university and bit stuck on the testing phase. How would you carry out testing for a system like this? any reference you can point me to?Thanks for the help!
Hello sir, i am little confused about the calib_data used in this video please any one can explain..., is that aruco marker is already calibrated or he estimated the distance without aruco marker calibration
regarding distance calculation, how much the accuracy improve by ?? im still getting like -50 mm error in distance calculation by still using the sqrt(x^2 + y^2 + z^2)
The accuracy may decrease as you go farther away from the camera 📷, 5cm is too much make sure to input an accurate measurement of the marker size, that will increase the accuracy.
Thanks for the prompt reply, could this be due to the cameras optical center not being exactly at origin or that shouldn’t matter? Thanks Just trying to rule out possibilities
It has been verified, that z is the distance you can check out the short video on Channel, where I have used a scale to measure the distance in real time.
@@AiPhile It will be approximately correct because when the marker is near to the camera... x and y you can not vary too much (marker will go out of the frame) but if you are far from camera... space for varying x and y will be much larger and it will be significant so that you need to take direct distance of aruco marker frame instead of just z
I have done Lots of experiment on it, Although I am facing a problem regarding constant percentage error offset for measuring z coordinate... i have created a set up to measure difference between actual z and opencv measured z both are having much larger difference on my laptop camera... It would be very helpful If you can help me on it
Basically we are getting rotation vector and translation vector, as you see in the code for x y and z you can calculate the yaw pitch and roll out of that.
@@AiPhile I tried doing that but the values were not changing upon rotating the aruco marker, so I searched and on the internet, it said that the rotation matrix is a mathematical way of representing angles in openCV but would have to be converted to Euler angles for the orientation(roll, pitch and yaw). How can we do that? Also thank you so much for the prompt reply.
Hi ... I have followed your AR course on YT and need help. By applying your code with my 4k IP cam (also bigger chessboard & markers), I couldn't estimate the distance of markers accurately. i.e. a distance 3.5 m vs calculated value of 4.99-5.10 m. Would you kindly help me fix this..
Thanks ever so much! You made my day
You are so welcome!
It was really helpful. Thank you so much!!!
Thank you so much for your course!
My pleasure!
need help!
during aruco detection it gives wrong rvec values. the marker has 0,0,0 (roll pitch yaw) whereas the output of the rvec value after detection gives -1.9,-1.9,-0.5.
rvecs, tvecs, _ = aruco.estimatePoseSingleMarkers(corners, marker_size, k, distortion_coefficients)
ros gazebo simulation (distortion coefficients all 0, marker size - 2cm, camera matrix -f_x & 0 & c_x \\
0 & f_y & c_y \\
0 & 0 & 1
I double checked and it is perfect now! thankyou so much. I am doing this for a project at university and bit stuck on the testing phase. How would you carry out testing for a system like this? any reference you can point me to?Thanks for the help!
Check out blog post of realpython
Hello sir, i am little confused about the calib_data used in this video please any one can explain..., is that aruco marker is already calibrated or he estimated the distance without aruco marker calibration
no you need to calibrate the camera first, please check out the source code on GitHub repository.
regarding distance calculation, how much the accuracy improve by ?? im still getting like -50 mm error in distance calculation by still using the sqrt(x^2 + y^2 + z^2)
The accuracy may decrease as you go farther away from the camera 📷, 5cm is too much make sure to input an accurate measurement of the marker size, that will increase the accuracy.
Thanks for the prompt reply, could this be due to the cameras optical center not being exactly at origin or that shouldn’t matter?
Thanks
Just trying to rule out possibilities
@@preenitkshirsagar7351 once you have properly calibrated the camera these things won't effect much, I would recommend to double check that.
@@AiPhile great! thankyou so much i will double check
Different marker sizes show z value different distances. How can I adjust it ???
You have to provide actual size of marker, to estimate the distance.
Can we find angles using fiducial markers
Distance is calculated wrongly, You can not take z as distnce from camera... It should be sqrt(x^2 + y^2 + z^2) where xyz are parameters of tvec
It has been verified, that z is the distance you can check out the short video on Channel, where I have used a scale to measure the distance in real time.
Here is the video ua-cam.com/users/shortsgBxm-WscmgU
@@AiPhile It will be approximately correct because when the marker is near to the camera... x and y you can not vary too much (marker will go out of the frame) but if you are far from camera... space for varying x and y will be much larger and it will be significant so that you need to take direct distance of aruco marker frame instead of just z
I have done Lots of experiment on it, Although I am facing a problem regarding constant percentage error offset for measuring z coordinate... i have created a set up to measure difference between actual z and opencv measured z both are having much larger difference on my laptop camera... It would be very helpful If you can help me on it
Let me me verify, then I will change this accordingly, thank you so man❤
Hello, I noticed there are no estimates for roll, pitch and yaw of the aruco. How can we get that data?
Basically we are getting rotation vector and translation vector, as you see in the code for x y and z you can calculate the yaw pitch and roll out of that.
@@AiPhile I tried doing that but the values were not changing upon rotating the aruco marker, so I searched and on the internet, it said that the rotation matrix is a mathematical way of representing angles in openCV but would have to be converted to Euler angles for the orientation(roll, pitch and yaw). How can we do that? Also thank you so much for the prompt reply.
Text me on Instagram
Is there a tutorial on how to do world tracking with opencv?
?
Hi ... I have followed your AR course on YT and need help. By applying your code with my 4k IP cam (also bigger chessboard & markers), I couldn't estimate the distance of markers accurately. i.e. a distance 3.5 m vs calculated value of 4.99-5.10 m.
Would you kindly help me fix this..
Have you provided the actual size of marker?
@@AiPhile yes, I have.it was between the edge of the black area. Was I correct?
@@ohm6403 yeah
m not an expert but I think u should check if you have calibrated your camera properly.
You are right man.
Hello, can the distance be used as a depth estimation of the aruco? To replace a depth camera reading, for example
You are right.
I am getting this error ' cv.2 aruco doesn't have attributes dictionary _get'. How can I correct it
Install mentioned version of opencv and clone the repository use the code
hai, do you have the source code in c++?
you can easily find this on GitHub.
what is marker size is 8 centimeter
You can measure the size of marker and adjust accordingly in the script.
@@AiPhile you say ,i need to measure the physical length and breadth of thr aruco marker??
Yeah, just measure one cause markers are in square ⬛ size.
@@AiPhile okay thanks for your rply