Matthew L Walter

US Patent:

20230080638, Mar 16, 2023

Filed:

Mar 11, 2022

Appl. No.:

17/692357

Inventors:

- Los Altos CA, US
Adrien David Gaidon - Mountain View CA, US
Rares A. Ambrus - San Francisco CA, US
Igor Vasiljevic - Fort Lauderdale FL, US
Jiading Fang - Chicago IL, US
Gregory Shakhnarovich - Chicago IL, US
Matthew R. Walter - Chicago IL, US

Assignee:

Toyota Research Institute, Inc. - Los Altos CA
Toyota Technological Institute at Chicago - Chicago IL

International Classification:

G06T 7/80
G06T 7/50
G06T 3/00
G06T 5/00
H04N 17/00

Abstract:

Systems and methods described herein relate to self-supervised learning of camera intrinsic parameters from a sequence of images. One embodiment produces a depth map from a current image frame captured by a camera; generates a point cloud from the depth map using a differentiable unprojection operation; produces a camera pose estimate from the current image frame and a context image frame; produces a warped point cloud based on the camera pose estimate; generates a warped image frame from the warped point cloud using a differentiable projection operation; compares the warped image frame with the context image frame to produce a self-supervised photometric loss; updates a set of estimated camera intrinsic parameters on a per-image-sequence basis using one or more gradients from the self-supervised photometric loss; and generates, based on a converged set of learned camera intrinsic parameters, a rectified image frame from an image frame captured by the camera.