Invention Grant
- Patent Title: Fully convolutional color constancy with confidence weighted pooling
-
Application No.: US15649950Application Date: 2017-07-14
-
Publication No.: US10713816B2Publication Date: 2020-07-14
- Inventor: Yuanming Hu , Baoyuan Wang , Stephen S. Lin
- Applicant: Microsoft Technology Licensing, LLC
- Applicant Address: US WA Redmond
- Assignee: Microsoft Technology Licensing, LLC
- Current Assignee: Microsoft Technology Licensing, LLC
- Current Assignee Address: US WA Redmond
- Agency: Schwegman Lundberg & Woessner, P.A.
- Main IPC: G06T7/90
- IPC: G06T7/90 ; H04N1/60 ; G06T5/00 ; G06N3/08 ; G06N3/04

Abstract:
Disclosed in some examples, are methods, systems, and machine readable mediums that correct image color casts by utilizing a fully convolutional network (FCN), where the patches in an input image may differ in influence over the color constancy estimation. This influence is formulated as a confidence weight that reflects the value of a patch for inferring the illumination color. The confidence weights are integrated into a novel pooling layer where they are applied to local patch estimates in determining a global color constancy result.
Public/Granted literature
- US20190019311A1 FULLY CONVOLUTIONAL COLOR CONSTANCY WITH CONFIDENCE WEIGHTED POOLING Public/Granted day:2019-01-17
Information query