temugeb / python_stereo_camera_calibrate Goto Github PK

After stereo calibration, I tried to run SGBM algorithm which compute the difference between the images taken by two cameras. Before running SGBM, there is a necessary step called image rectify with stereo parameters. The code was like below


I tested the code with the images in your synched folder


but the rectified images was like this


Obviously it seems not good. But the stereo parameters are obtained through these images. How the rectified images just show part of the original ones.

Hi! A big thank for your work, very nice and helpful!

I'm trying to move my custom avatar with some generated keypoints!
What I first did was to take your raw data and sen it to my avatar to see what it is doing!
Here you can see the first frame of the avatar moving only the left shoulder

So it is kind of strange!
I know it was a fast try, but may you help to understand where I have to change your code?
Is it due to the Rotation?

The avatar is on the Threejs coordinate system !

Thank you

I try using the example program with the images you provide on here. However, I cannot detect the checker board at all. Can you give me some idea what should be the cause?

btw, thank you for the awesome program and articles.

Hi, I captured images using 2 webcams of the same specifications simultaneously. rmse value returned when each camera is calibrated is around 0.11.
When they are setreocalibrated, the rmse value returned is around 66.

Is it a good rmse value? if not how to achieve good rmse value?

Now I gone through the blog post, depth of selected points are correct, but what if I want the depth map of the frame.

Hey, great repository, but all mentions of rows and columns seem to be switched in the code. Also in the readme you should tell people to enter distance between squares in mm so then the output matrices end up being in mm .

Hi, I was trying 3D pose estimation using $3$ cameras: i.e., first finding 2D keypoints from each of the 03 camera frames and using cv2.triangulatePoints (to triangulate and get the 3D human pose). In such a case, how to get the camera extrinsic parameters?

Specifically, suppose for $3$ cameras, should we use $[C1, C2]$, $[C2, C3]$ pairs or should we use $[C1, C2]$, $[C1, C3]$ to find the extrinsic matrix of the cameras?

Hi,

I was wondering if there is a way to easily adapt this for pre-recorded videos instead of live? I put the video paths in the yaml instead of a camera id number but the issue is that the video finishes and then it still expects a constant input feed. If anyone has made this work please let me know!

Thanks

Hi Temuge,
Thank you for your nice work and sharing. I read your post about camera calibration in detail and found your link to calibration frames didn't work. Could you please post them again so I can download? That will be very helpful. I will appreciate your any response.

Thanks,
Rui

Hi, I follow the steps and got rmse value around 1. However when I rebuilt my chessboard to 3D figure, it became parallelogram not rectangle.
I had a 5 x 4 chessboard with 90mm world scale, the rebuilt figure got 455mm and 356mm borders which are pretty close. But the angles are 35 and 55, they should be 90.

Is there anything I should do before transforming the triangulation xyz points to 3D figure?

Hi Sir

Thanks for your code !
a question what is world_scaling refer to ?

Thanks !

DLT supoort 2 points to triangulate 3d point

if i can provide more than 2 points like 3 or 4, is there any method to do that ?

Thanks in advanced !

HI. Sorry to disturb you.

I am wanting to use your stereo camera calibration code in my final year project am running into a problem.
I am currently building a game that you play with a ball, I have 2 webcams that track the ball and let me know when the ball has hit the wall. this works fine but I am having problems with the optional part. I want the wall to be set as the Z = 0 plain.

I have changed your code a bit(sorry) so that it takes a new photo to calibrate to the wall, this is done by a projector as I need it to be calibrated to the cameras. the only thing different is the size of the checkbox.

The funny and irritating part is that I got my code to work but then I went and recalibrated my cameras to get a better shot and now they are not working (And I have NO idea why). Could you please help me even if it is just for 5 min I will be very grateful.
https://github.com/Jonathan5320/EEE4022-Final-year-project.git

arigatōgozaimasu

Hi,

I tried bodypose3d using 2 webcams now I want to explore with real cameras or mobile phones.
Do you have any idea how to capture videos from multiple phones or cameras at the same time using any Application or IoT devices?

im getting this error:
cv2.error: OpenCV(4.6.0) D:\a\opencv-python\opencv-python\opencv\modules\calib3d\src\calibration.cpp:3694: error: (-215:Assertion failed) nimages > 0 in function 'cv::calibrateCameraRO'

i found out that findChessboardCorners isnt able to find my checkerboard
how do i fix this?

Hi, can you please share the images you used for single camera calibration and also synchronized frames?

Hi can you please provide some calibration images you collected so that I can reproduce results you got once? With images I collected, the reprojection error is not going below five.