Gustavo Carneiro

Gustavo Carneiro is a computer scientist who applies deep learning to medical image analysis and computer vision. After a PhD at the University of Toronto in 2004, he worked at Siemens Corporate Research and as a Marie Curie fellow in Lisbon, then joined the University of Adelaide in 2011, where he became a professor in 2019 and directed Medical Machine Learning at the Australian Institute for Machine Learning. In 2016 he coauthored a network that learned single-view depth without annotated depths, and in 2019 a study of hidden stratification in medical imaging. His move to the University of Surrey came in December 2022; in 2023 he was one of the Chief Investigators of the IMAGENDO endometriosis-diagnosis system, and in 2024 he published the textbook Machine Learning with Noisy Labels. At Surrey, where he is a Professor of AI and Machine Learning, recent work covers future knee X-ray prediction for osteoarthritis and AI systems that know when to defer to human pathologists.

Now at: University of Surrey
Badges: Bridge
Wayfarer

Timeline

1996–2004

A Toronto computer science PhD

He earns a BSc at the Universidade Federal do Rio de Janeiro in 1996 and an MSc at the Instituto Militar de Engenharia in 1999. In 2004, he completes a PhD in computer science at the University of Toronto under Allan Jepson, with a dissertation on phase-based local image descriptors.
2004–2011

Siemens and a Marie Curie fellowship

He does postdoctoral work with Nuno Vasconcelos at the Statistical Visual Computing Lab of UC San Diego in 2004 and, as an NSERC fellow, with David Lowe at the University of British Columbia in 2005. From 2006 to 2008 he is a research scientist at Siemens Corporate Research in Princeton, and from 2008 to 2011 he is a Marie Curie fellow and visiting assistant professor at Instituto Superior Técnico in Lisbon within the CMU-Portugal program.
2011–2022

Professor at Adelaide

He joins the University of Adelaide as a senior lecturer in 2011, becomes an associate professor in 2015 and a professor in 2019. From 2018 until 2022 he is Director of Medical Machine Learning at the Australian Institute for Machine Learning, and he is an ARC Future Fellow from 2020.
2016

Depth estimation without labels

In 2016, he coauthors an ECCV paper with Ravi Garg, Vijay Kumar BG and Ian Reid that trains a convolutional network to predict single-view depth without annotated ground truth depths. Trained on less than half of the KITTI dataset, it gives performance comparable to state-of-the-art supervised methods.
2019

Hidden stratification in medical imaging

In 2019, he coauthors an ML4H at NeurIPS extended abstract with Luke Oakden-Rayner, Jared Dunnmon and Christopher Ré. It names hidden stratification, where a medical imaging model performs poorly on unidentified subsets of a population, and finds relative performance differences of over 20% on clinically important subsets.
2022–2023

IMAGENDO for endometriosis diagnosis

He moves to the University of Surrey in December 2022 as Professor of AI and Machine Learning. In August 2023, Surrey presents IMAGENDO, an AI system that analyses ultrasound and MRI data to shorten the diagnosis of endometriosis, a disease that averages more than six years to diagnose correctly; he is one of its Chief Investigators, and development is led by the Robinson Research Institute at the University of Adelaide.
2024

A textbook on noisy labels

In February 2024, Elsevier publishes his book Machine Learning with Noisy Labels: Definitions, Theory, Techniques and Solutions. It defines the types of label noise, introduces the theory behind the problem and presents the main techniques for making effective use of noisy-label training sets.
As of October 2026

Professor of AI at Surrey

He is Professor of AI and Machine Learning and a PAI Fellow at the University of Surrey. In October 2025, Surrey announces a MICCAI 2025 study in which a diffusion model generates a "future" version of a patient's knee X-ray to forecast osteoarthritis progression. In March 2026, Surrey and Monash University researchers, including him, present a method that teaches AI to defer to human pathologists in cancer diagnosis.

Career

Since 2022University of SurreyProfessor of AI and Machine Learning
2018–2022Australian Institute for Machine LearningDirector of Medical Machine Learning
2011–2022University of AdelaideSenior lecturer, Associate professor
2008–2011Instituto Superior TécnicoMarie Curie fellow and visiting assistant professor
2006–2008SiemensResearch scientist
2005University of British ColumbiaNSERC fellow
2004University of California, San DiegoPostdoctoral researcher