OpenCV Python Depth Map Stereo Vision

In this video, I will go over depth maps in OpenCV using Python in VS Code. I will explain what depth maps are and how to calculate and tune the parameters to get good depth map results. We will take two stereo images from the Middlebury dataset and use the block matching and semi-global matching algorithms and compare the depth map results.
0:00 Introduction
0:23 What is depth map?
1:19 Why do we need depth map?
1:35 How does depth map work?
3:56 Code
Thanks for watching! If you found this video helpful, please like, subscribe and share:
/ @kevinwoodrobotics
Social:
LinkedIn: / kevinwoodrobotics
Github: github.com/kevinwoodrobotics
Instagram: / kevinwoodrobotics
Twitter: / kevinwoodrobotics

[@kevinwoodrobotics]

Code and Doc: kevinwoodrobotics.com/product/opencv-python-depth-map/
OpenCV Python Playlist Code and Doc: kevinwoodrobotics.com/product/opencv-python-tutorials-full-playlist/
Learn more about camera calibration: OpenCV Python Camera Calibration th-cam.com/video/H5qbRTikxI4/w-d-xo.html

[@jeevithaglory8557]

Useful content ❤

[@realblythe]

Great topic. Thank you for sharing

[@kevinwoodrobotics]

Thanks for tuning in

[@mrbentv1762]

great vlog❤️watching this video

[@kevinwoodrobotics]

Thanks!

[@jasminecadizal4541]

Thank you for sharing 😊❤

[@haidarmalik0]

Great video.

[@kevinwoodrobotics]

Thanks!

[@mrbentv1762]

good job👏

[@joymadula]

Nice video!

[@kevinwoodrobotics]

Thanks!

[@user-pu5dv2ly3o]

Thank you so much! Have used the ssd or normalized coefficients correlation in your code? If yes, how did you?

[@ladislaus86]

Awesome

[@kevinwoodrobotics]

Thanks!

[@Famousclip0]

Good work

[@kevinwoodrobotics]

Thanks!

[@Famousclip0]

@@kevinwoodrobotics my pleasure

[@analynbalanlayos2519]

nice

[@user-pu5dv2ly3o]

Thank you so much! Have you used the ssd or normalised coefficients correlation in your code? If yes, how did you?

[@kevinwoodrobotics]

Thanks for tuning in! The stereo.compute handles that part for you so I’m not actually implementing that part for the stereo matching. But I talk about it in my template matching video.

[@kevinwoodrobotics]

github.com/kevinwoodrobot/opencv-python-tutorials/tree/main/38-template-matching

[@kevinwoodrobotics]

th-cam.com/video/BNXu20ToDl4/w-d-xo.htmlsi=lo35MiqcqQ8e7fVb

[@realblythe]

First notify here

[@georgepal1215]

What is the baseline and f used for this disparity? Where in the code we should define these parameters of our configuration?

[@kevinwoodrobotics]

Great question. You’ll need it when you try to compute the actual depth value z from the disparity, typically for 3D reconstruction or distance measurements.

[@jak-zee]

Thanks for the video. Would this work if one of the view is a fisheye camera from a different view point?

[@kevinwoodrobotics]

It’s better to calibrate your camera and undistort your images first. I have a video on camera calibration if you want to check it out!

[@jak-zee]

Thanks @@kevinwoodrobotics . In case you want some ideas for your future videos, you can do depth map on fisheye cameras. Or automatic calibration without a cherckerboard. Good luck! Keep it up.

[@karen3541]

still watching 🙃

[@anidea8012]

is it better than ODM, COLMAP, NERF or NERFSTUDIO ?

[@kevinwoodrobotics]

There’s been a lot of new work on NERF lately with great results, however it really depends on the scene you are looking at. COLMAP is typically for sparse point clouds, whereas SGM is for dense. SGM will only give you point cloud for a given viewpoint so you’ll have to stitch them together yourself.