Denoising Monte Carlo renderings using machine learning with importance sampling
Supervised machine learning using neural networks is applied to denoising images
rendered by MC path tracing. Specialization of neural networks may be achieved by using a …
rendered by MC path tracing. Specialization of neural networks may be achieved by using a …
Temporal techniques of denoising Monte Carlo renderings using neural networks
A modular architecture is provided for denoising Monte Carlo renderings using neural
networks. The temporal approach extracts and combines feature representations from …
networks. The temporal approach extracts and combines feature representations from …
Denoising Monte Carlo renderings using progressive neural networks
Supervised machine learning using neural networks is applied to denoising images
rendered by MC path tracing. Specialization of neural networks may be achieved by using a …
rendered by MC path tracing. Specialization of neural networks may be achieved by using a …
Burst image fusion and denoising using end-to-end deep neural networks
C Chen, GM Larsson, J Liu, S Sridhar… - US Patent …, 2023 - Google Patents
Electronic devices, methods, and non-transitory program storage devices for leveraging
machine learning to perform improved image fusion and/or noise reduction are disclosed …
machine learning to perform improved image fusion and/or noise reduction are disclosed …
Action recognition method and apparatus, computer storage medium, and computer device
LUO Donghao, Y Wang, GUO Chenyang… - US Patent …, 2024 - Google Patents
An action recognition method includes: obtaining original feature submaps of each of
temporal frames on a plurality of convolutional channels by using a multi-channel …
temporal frames on a plurality of convolutional channels by using a multi-channel …
Statistical model training systems
Subsets of training data are selected for iterations of a statistical model through a training
process. The selection can reduce the amount of data to be processed by selecting the …
process. The selection can reduce the amount of data to be processed by selecting the …
Denoising filter
S Csefalvay - US Patent 11,915,396, 2024 - Google Patents
A pixel filter has a filter module that performs a first recursive filter operation in a first
direction through a sequence of pixels to form a first filtered pixel value for each pixel, and …
direction through a sequence of pixels to form a first filtered pixel value for each pixel, and …
Temporal techniques of denoising monte carlo renderings using neural networks
2023-02-08 Assigned to THE WALT DISNEY COMPANY (SWITZERLAND) GMBH
reassignment THE WALT DISNEY COMPANY (SWITZERLAND) GMBH ASSIGNMENT OF …
reassignment THE WALT DISNEY COMPANY (SWITZERLAND) GMBH ASSIGNMENT OF …
Simultaneous dereverberation and denoising via low latency deep learning
Y Ji, R Chen, Z Liu, X Zhao, S Yao - US Patent 11,404,055, 2022 - Google Patents
A method includes extracting, from a frame of the audio sample, spectral features indicative
of cochlear impulse response of an auditory system; obtaining an estimate of a speech …
of cochlear impulse response of an auditory system; obtaining an estimate of a speech …
Denoising filter
S Csefalvay - US Patent 11,386,528, 2022 - Google Patents
A pixel filter has a filter module that performs a first recursive filter operation in a first
direction through a sequence of pixels to form a first filtered pixel value for each pixel, and …
direction through a sequence of pixels to form a first filtered pixel value for each pixel, and …