Mapping the 3D World to an Image - 5 Minutes with Cyrill

preview_player
Показать описание
Mapping 3D points to 2D pixel locations explained in 5 minutes
Series: 5 Minutes with Cyrill
Cyrill Stachniss, 2021

Credits:
Video by Cyrill Stachniss
Special thanks to Olga Vysotska and Igor Bogoslavskyi
Intro music by The Brothers Records
Рекомендации по теме
Комментарии
Автор

Remarkable how a complicated process can be covered so clearly and concisely in just 5 minutes!!

willw
Автор

0:51 1:16 4 coordinate systems 1:56 camera location 3:48 formula interpretation 4:06 not invertible b/c info loss from 3D to 2D (given a 3D point, we can use the calculated P to get the corresponding 2D point, but given a 2D point, we cannot use P to get the corresponding 3D point b/c loss of info) 4:44 can partially invert with a 1D solution space 4:51 5:07 5:13

willw
Автор

Your videos have been the best tool to clear my computer vision concepts. Thank you so much :) !!

ashleshagill
Автор

Thanks! I really appreciate your lectures.

goldenshale
Автор

Excellent. I feel much more comfortable with the concept, and have the correct terminology for further study.

atrowell
Автор

Fast and informative. Thank you so much

imadelbouazzaoui
Автор

I have a question, maybe you can give me a hint how I have to continue. I use a camera and can detect my aruco marker. The camera is calibrated and I have the camera matrix (intrinsics), distortion coefficients and the rotation & translation vector (extrinsic). And by the aruco functions I can detect the center of my aruco marker in the given image. But HOW can I caluclate the x, y coordinates of the marker in real world coordinates by the given parameters??? I dont get it :D I want to now when I make pic1, than move the marker, do pic2, I want to know how many mm in the real world the aruco marker moved in the x-y layer. And the distance to the object is not constant. But I do not want to know the distance. I just want to get accurate x, y coordinates. Thanks for your great video collection by the way!

ericbighead
Автор

Thank you so much for this amazing content.Would you like to make a video about visual odometry using single camera?

marzanalam
Автор

Thank you. Could you talk about what are the common feature extractors and descriptors for 3D point cloud (with and without RGB)?

yousofebneddin
Автор

Please make a video on Feedback Particle filter

roy