I recorded a video for a 10x7 checkerboard and tried to extract the intrinsic calibration for the camera.
Not only is the fps very low (0.5-2fps) but also the tool does not manage to find any constraints at all.
The command I use:
./calibration-intrinsics-constraints --use-checkerboard=10x7 --input-video=/home/nvidia/Dev/CameraCalibration/intrinsics/ixxx_front_left.h264 --output=/home/nvidia/Dev/CameraCalibration/intrinsics/camera-left.json
Any ideas what could be the problem?
This is a calibration tool so it may takes time to have an accurate result.
Could you attach an image of your checkboard pattern with us?
Please noticed that the format should be [height]x[width].
The value is the number of intersection points, which should be square number-1.
Here is a picture.
Is it necessary for the calibration to always see the full checkerboard in the image?
I let it run for about 5 minutes without even one constraint being found (5 min at 1fps that is)
It looks like your pattern is 6x9.
Could you give it a try?
./calibration-intrinsics-constraints --use-checkerboard=6x9 ...