Join My AI Career Program 👉 www.nicolai-nielsen.com/aicareer Enroll in the Investing Course outside the AI career program 👉 nicolai-nielsen-s-school.teachable.com/p/investment-course Camera Calibration Software and High Precision Calibration Boards 👉 camera-calibrator.com/
Yeah charuco is the most accurate by far. Circle pattern is only used for very low res cameras but is kinda outdated. Yeah you can use the stereo calibration and rectification and you will get accurate depth info down to cm acc
Is it possible to use a ChArUco board for image stitching, as opposed to camera calibration? That is, use the ChArUco board as a kind of reference image in images from 3 different cameras and stitch these images together based on that.
Yes that should work, you need three features to get a camera pose so the like 30 features on the board should be plenty. Will obviously be better if the board is large compared to the background. SIFT should be able to do this anyway unless youre images are pretty blank / obscured
Join My AI Career Program
👉 www.nicolai-nielsen.com/aicareer
Enroll in the Investing Course outside the AI career program
👉 nicolai-nielsen-s-school.teachable.com/p/investment-course
Camera Calibration Software and High Precision Calibration Boards
👉 camera-calibrator.com/
Tutorial for classification custom dataset and with webcam please sir.
Multi-camera support would be nice.
Will be added for sure! Mainly been getting requests for mono and stereo
is using charuco board or circular pattern is better than chessboard pattern also were you you able to get accurale depth using streo camera
Yeah charuco is the most accurate by far. Circle pattern is only used for very low res cameras but is kinda outdated. Yeah you can use the stereo calibration and rectification and you will get accurate depth info down to cm acc
Is it possible to use a ChArUco board for image stitching, as opposed to camera calibration? That is, use the ChArUco board as a kind of reference image in images from 3 different cameras and stitch these images together based on that.
Yes that should work, you need three features to get a camera pose so the like 30 features on the board should be plenty. Will obviously be better if the board is large compared to the background. SIFT should be able to do this anyway unless youre images are pretty blank / obscured