Patent · US Active

Universal correspondence network

US10115032B2 · kind B2 · utility

4Cited by

0References

15Claims

0Family size

Assignee

NEC CORPORATION · JP

Inventors

Manmohan Chandraker · Santa Clara, US
Christopher Bongsoo Choy · Los Angeles, US
Silvio Savarese · Palo Alto, US

Key dates

Filing date	Nov 3, 2016
Grant date	Oct 30, 2018
Priority date	—
Expiry date	Feb 7, 2037

Classification

Technology area (CPC G)Physics
CPC primaryG06T2207/20084
WIPO fieldComputer technology
WIPO sectorElectrical engineering

Abstract

A computer-implemented method for training a convolutional neural network (CNN) is presented. The method includes extracting coordinates of corresponding points in the first and second locations, identifying positive points in the first and second locations, identifying negative points in the first and second locations, training features that correspond to positive points of the first and second locations to move closer to each other, and training features that correspond to negative points in the first and second locations to move away from each other.

Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.