01In progress
Two viewsStereo vision lab
Depth from disparity. Move the cameras, change the baseline, and watch Z = f·B/d respond. Then learn to recover depth from the two images.
Open the project→Computer vision
Where is this thing?
Three projects on one question, with different sensors. Two cameras give depth from the shift between two images. Many cameras give a 3D scene. A microphone array answers the same question with sound. They share the same geometry, so they are best read together.
The series
02In progress
Many views3D reconstruction from images
The same epipolar geometry across many photos: camera poses, a sparse point cloud, then a scene you can walk around.
Open the project→03Planned
Another sensorSound source localization
The delay between microphones plays the part of disparity. Same question, different physics.
Open the project→Why they belong together
Each project starts from the classical geometry and then adds a learned model, so the two can be compared.