Michael Hobley
Michael Hobley, Ph.D.
Postdoctoral Reasearcher @ CalTech
ML Product Owner @ Noonien

I am currently a postdoctoral researcher at CalTech working in AI for ecology. I am also the founder of Noonien, a Machine Learning consultancy engineering custom solutions for businesses and indivuals.
I am a recent DPhil graduate in Computer Vision and Machine Learning from the Active Vision Laboratory group at the Engineering Science department of the University of Oxford.
My research interests and experience are broad, spanning: ecology, weakly-supervised computer vision, large-scale data understanding, generative AI, LLMs, and engineering pipeline acceleration.

News

08/01/2025
Started postdoctoral reasearch position at CalTech focusing on AI for ecology.
01/10/2024
Presented ABC123 as an Oral at ECCV Milan.
05/06/2024
EngineBench announced, code and dataset released.
21/05/2024
Noonien Ltd. Founded to deliver bespoke machine learning solutions to business and individual clients. Noonien's initial contracts are with state of the art engineering research and development companies.
19/05/2024
ABC Easy as 123 code release and MCAC Dataset release.
09/05/2024
Presented PoRF: Pose Residual Field for Accurate Neural Surface Reconstruction at ICLR.
29/04/2024
Completed DPhil at Oxford. 'Counting with Limited Supervision' was supervised by Victor Prisacariu and was examined by Minh Hoai Nguyen and Ioannis Havoutis.
18/06/2023
Presented Learning to Count Anything: Reference-less Class-agnostic Counting with Weak Supervision at CVPR.
01/07/2022
Learning to Count Anything code release and FSC-133 dataset release.
03/12/2018
Presented Say Yes to the Dress: Shape and Style Transfer Using Conditional GANs at ACCV.