Matthew Alun Brown

I am a Principal Scientist at Wayve.

Before that, I was a Research Scientist at Google DeepMind / Google Research, an Associate Professor (Reader) at the University of Bath (2011-2015), Founder and CTO of Cloudburst Research Inc. (2009-2015), and a Postdoctoral Researcher at EPFL (2009-2011), UBC (2008-2009) and Microsoft Research (2006-2007).

My interests are in Computer Vision, Machine Learning and Environmental Informatics. In 2017 I won the ICCV Helmholtz Prize (test of time award) for my work with David Lowe on panoramic image stitching.

Projects

OmniNOCS: A unified NOCS dataset and model for 3D lifting of 2D objects

OmniNOCS: A unified NOCS dataset and model for 3D lifting of 2D objects

ECCV, 2024

Large-scale Normalized Object Coordinate Space (NOCS) dataset with 90+ object classes across different domains, and a novel, transformer-based monocular NOCS prediction model.

Module-wise Adaptive Distillation for Multimodal Foundation Models

Module-wise Adaptive Distillation for Multimodal Foundation Models

Neurips, 2023

Multi-armed bandits for adaptive layerwise distillation of large pre-trained foundation models.

FiG-NeRF: Figure Ground Neural Radiance Fields for 3D Object Category Modelling

FiG-NeRF: Figure Ground Neural Radiance Fields for 3D Object Category Modelling

3DV, 2021

Given a lot of images of an object category, you can train a NeRF to render them from novel views and interpolate between different instances.

Mobile Video Networks for Efficient Video Recognition

Mobile Video Networks for Efficient Video Recognition

Dan Kondratyuk, Liangzhe Yuan, Yandong Li, Li Zhang, Mingxing Tan, Matthew Brown, Boqing Gong
CVPR, 2021

MoViNets are a family of efficient video classification models supporting frame-by-frame inference on streaming video.

Rethinking Class-Balanced Methods for Long-Tailed Visual Recognition from a Domain Adaptation Perspective

Rethinking Class-Balanced Methods for Long-Tailed Visual Recognition from a Domain Adaptation Perspective

CVPR, 2020

Augment classic class-balanced learning by estimating differences between class-conditioned distributions via meta-learning

Federated Visual Classification with Real-World Data Distribution

Federated Visual Classification with Real-World Data Distribution

Tzu-Ming Harry Hsu, Hang Qi, Matthew Brown
ECCV, 2020

Create datasets and study the effect of non-identical data distribution on Federated visual classification.

Measuring the Effects of Non-Identical Data Distribution for Federated Visual Classification

Measuring the Effects of Non-Identical Data Distribution for Federated Visual Classification

Tzu-Ming Harry Hsu, Hang Qi, Matthew Brown
Neurips FL Workshop, 2019

Initial work on non-IID data distribution for visual classification via Federated Learning

Low Shot Learning wtih Imprinted Weights

Low Shot Learning wtih Imprinted Weights

Hang Qi, David Lowe, Matthew Brown
CVPR, 2018

Recognise novel categories by imprinting final layer weights using the activations from one or more exemplars.

Frame-recurrent video super-resolution

Frame-recurrent video super-resolution

CVPR, 2018

End-to-end trainable frame-recurrent video super-resolution framework that uses the previously inferred HR estimate to super-resolve the subsequent frame

Learning to Segment via Cut-and-Paste

Learning to Segment via Cut-and-Paste

Tal Remez, Jonathan Huang, Matthew Brown
ECCV, 2018

Learn segmentation without ground truth by playing a game of cut and paste in an adversarial learning setup

Unsupervised Learning of Depth and Ego-Motion from Video

Unsupervised Learning of Depth and Ego-Motion from Video

CVPR, 2017

SFMLearner: Joint learning of monocular depth plus camera pose from image streams without camera or depth supervision.

Kernel Regression for Real-Time Building Energy Analysis

Kernel Regression for Real-Time Building Energy Analysis

Matthew Brown, Christopher Barrington-Leigh, Zosia Brown
Journal of Building Performance Simulation, 2012

Building energy modelling and event detection using kernel regression

Multi-spectral SIFT for Scene Category Recognition

Multi-spectral SIFT for Scene Category Recognition

Matthew Brown, Sabine Süsstrunk
CVPR, 2011

New RGB-NIR scene dataset and multispectral SIFT variant

Discriminative learning of local image descriptors

Discriminative learning of local image descriptors

Matthew Brown, Gang Hua, Simon Winder
PAMI, 2010

Learn local descriptors by sampling corresponding image patches from large scale 3D reconstructions.

City-Scale Location Recognition

City-Scale Location Recognition

Grant Schindler, Matthew Brown, Rick Szeliski
CVPR, 2007

Location recognition in large image sets using informative feature vocabulary trees

Unsupervised 3D Object Recognition and Reconstruction in Unordered Datasets

Unsupervised 3D Object Recognition and Reconstruction in Unordered Datasets

Matthew Brown, David Lowe
3DIM, 2005

Fully automatic structure and motion for unordered image sets.

Automatic Panoramic Image Stitching using Invariant Features

Automatic Panoramic Image Stitching using Invariant Features

Matthew Brown, David Lowe
IJCV 2007

Improved version of the AutoStitch panorama stitcher

Recognising Panoramas

Recognising Panoramas

Matthew Brown, David Lowe
ICCV 2003 IEEE Test of Time Award

First solution for panoramic stitching without user input



Teaching

UBC CPSC 425 Computer Vision

University of British Columbia, 2022, 2023

This course provides an introduction to the fundamental principles and applications of computer vision.

UW CSE P576 Computer Vision

University of Washington 2020, 2021

A masters course in computer vision.