Computer Vision

In 2018, I took Dr. Hyun Soo Park's special topics course on multiview 3D geometry, learning computer vision techniques including image transformation, epipolar geometry, and structure from motion using bundle adjustment.

Primary language: MATLAB, C++/OpenGL

Estimating 3D Social Saliency

Using a series of images from the same frame, we want to estimate the interaction between each person by extracting the socially salient features of the image. Several people are reconstructed in 3D, and an estimate of their gaze direction is used to determine the socially salient features. This is the first step in an ongoing project.

5561_Final_Project_Final.pdf

In-depth analysis of method

zach chavis poster (1).pdf

Abbreviated version of final project paper

Image-Based Rendering

Using a series of photographs of an object, and after solving for the camera poses and prescriptions, and the model of the object, photographs are combined in an intelligent way to recover the texture of the object. Digital renderings were used to measure accuracy of the method, meaning models were given and dense reconstruction was not needed. A mock paper was written for project submission.

5980_Final_Project.pdf

Structure From Motion

Using multiple cell phone camera images, minimized reprojection error of point correspondences to solve for camera poses and 3D point locations of scene.