Intro2Robotics Lecture 15: Computer Vision, world frame to camera frame to pixel coordinates
Ғылым және технология
Computer vision is one of the most powerful sensing modalities that currently exist for robotics. This lecture introduces computer vision with a pinhole camera model, and explains how to convert from world frame coordinates to camera frame coordinates to pixel coordinates.
The handout for this is located in the worksheets folder of uofh-my.sharepoint.com/:f:/g/...
Example Videos from
• How to Transform a Roo... How to Transform a Room into a Camera Obscura
• Camera Calibration wit... Camera Calibration with MATLAB
• Folding 3D Structures ... Folding 3D Structures with a Cozmo Robot Swarm!
• Puppies Running Into C... Puppies running Into Cameras
Full Playlist "Intro to Robotics": • Intro2Robotics Lecture...
Пікірлер: 15
Great Content. Intuitive, concise, with examples. Revised my concepts quickly. Thank you.
@AaronBecker
3 жыл бұрын
Thanks @SUDHIR PRATAP YADAV (B14EE033) I looked at your channel and you've been doing a lot of robotics research. What kind of robots are you going to build?
@SudhirPratapYadav
3 жыл бұрын
@@AaronBecker I am currently working on UR5 and using deep learning (specifically RL) based vision and control schemes to control it. In short visuomotor control using DeepRL. My research interest is AI based autonomous robotics. Currently I am focusing on vision aspect of it. I guess every robotist final aim is same☺️. To build autonomous intelligent robot.
@SudhirPratapYadav
3 жыл бұрын
@@AaronBecker also about my channel. Yeah I keep a tab on videos and courses. I love learning and teaching.
@AaronBecker
3 жыл бұрын
@@SudhirPratapYadav Nice! We're doing a project right now with the UR5 and vision too, using surgical scopes. I hope you are successful.
@SudhirPratapYadav
3 жыл бұрын
@@AaronBecker i wish same to you
Thanks for this amazing video and for the story at the end💗
@AaronBecker
3 жыл бұрын
Bed time stories are fun
Great Video , how do I calculate object tracking area from the camera FOV in image plan ?
@AaronBecker
2 жыл бұрын
I'm unsure what you are asking for. To calculate the number of pixels in the image, you can use connected components analysis: kzread.info/dash/bejne/pp2XvMWlhbzfaJs.html
on 2b. negative k with negative camera coordinates should of had positive u,v values
@seasnek7024
Ай бұрын
Went and looked to see if anyone else caught this too.
Great video, can you help me? I am trying to calculate real worl coordinates from camera coordinates with python. I can't find any source document with this topic. I just find that Field of view = 2*tan^-1(h/2f) f= sensor size h = object size I have to transform camera coordinates to real coordinates. What can I do? I hope I explained what I want to do
@AaronBecker
Жыл бұрын
Start by measuring a few points in both the real world and in your camera frame. I think kzread.info/dash/bejne/iXmN2cutkpnKhNI.html will give you a start
@toygarkodik3247
Жыл бұрын
@@AaronBecker thank you