Projects
Course and personal projects in computer vision, generative models, and machine learning.
Neural Radiance Fields
Fit a neural field to a 2D image, then trained a NeRF from multi-view images to render novel views of a 3D scene.
Diffusion Models
Sampling and editing with a pretrained DDPM, then building and training a class-conditioned diffusion UNet from scratch.
Panorama Stitching
Homography-based image warping and mosaicing, with automatic feature detection, description, and matching.
Face Morphing and Facial Keypoint Detection
Correspondence-based morph sequences, the mean face of a population, and caricatures by extrapolation.
Hybrid Images & Multiresolution Blending
Filters and frequencies — edge detection, image sharpening, hybrid images, and Laplacian stack blending.
Colorizing the Prokudin-Gorskii Collection
Aligning the three glass-plate color channels with an image pyramid to reconstruct color photographs from the 1900s.