This project is no longer accepting applications. Subscribe to our newsletter to be notified of new projects!
Develop a comprehensive 3D vision pipeline, transforming 2D images into 3D point clouds using Python and computer vision techniques.
Embark on an exciting journey into 3D computer vision, where you'll learn to transform 2D images into fascinating 3D point clouds. This Build Project is designed to introduce you to the world of 3D vision, regardless of your programming background. Over eight interactive workshops, you'll construct a 3D vision pipeline step by step, with all the necessary code structures provided. You'll use Python and Jupyter notebooks, you’ll receive guidance if you're new to these. Each workshop will include prepared code segments where you'll fill in key functions, supported by built-in tests to verify your solutions. You'll explore concepts like camera models, image processing, and stereo vision through hands-on exercises. By the project's end, you'll have contributed to a working 3D reconstruction tool, gaining insights into computer vision, basic Python programming, and 3D data visualization. This project offers a gentle introduction to technologies used in exciting fields like robotics and augmented reality, providing you with a unique learning experience regardless of your starting point.
Explore the course overview, set up your development environment, and learn how to compile and use Jupyter Book for your project deliverable.
Dive into how cameras work, focusing on the OpenCV pinhole camera model and lens distortion. You'll program a distortion function to see these concepts in action.
Learn about undistortion and image remapping. You'll implement an undistort function and apply it to real images.
Explore how to project 3D points onto 2D images and solve the Perspective-n-Point (PnP) problem. You'll write part of a PnP solver function
Understand how to work with two cameras simultaneously and perform triangulation. You'll implement a triangulation function to reconstruct 3D points.
Discover the basics of stereo matching and its role in 3D vision. You'll implement a simple stereo matching algorithm.
Learn how to convert disparity maps to depth information and create basic point clouds. You'll write a function to generate a point cloud from disparity data.
Integrate all components into a complete 3D reconstruction pipeline. You'll present your final project, demonstrating your 3D vision tool and discussing your learning journey.
Get access to all of our Build projects, including this one, by creating your Build account!
Get started by submitting your application.
We'll notify you when projects reopen. In the meantime, you can explore our resources and learn more about our Fellows.