Biography
Martin started as a DPhil student in 2015 with funding from the EPSRC and Google. He was a member of Somerville College where he also gave tutorials for first and second year undergraduates during the first two years of his DPhil.
His research focused on the application of deep neural networks in robot perception and planning. Other areas of interest included: perception from 2D camera images and 3D laser data, connections between neural networks and information theory, and probabilistic generative modelling. Martin led the development of Vote3Deep, a novel method for efficiently detecting objects in 3D point clouds.
In 2017, Martin spent three months at Facebook AI Research in Paris. He was supervised by Ben Graham and worked on semantic segmentation of 3D point clouds, winning a 3D segmentation competition at ICCV.
During his undergraduate studies at the Engineering Department, Martin interned at the management consultancy firm McKinsey in Frankfurt and at the Indian technology company Infosys in Pune. He was also involved in a variety of technology-related projects, most notably as a co-organiser of StartupBus and as a team leader at the Oxford Microfinance Initiative.
Martin enjoys participating in hackathons – successes include winning the Apple Award at HackZurich 2014 and the first prize at HackTrain 2015.
Publications
ORI Publications
“GENESIS: Generative Scene Inference and Sampling with Object-Centric Latent Representations” – M Engelcke, AR Kosiorek, O Parker Jones, I Posner; International Conference on Learning Representations (ICLR), 2020
“On the Limitations of Representing Functions on Sets” – E Wagstaff, FB Fuchs, M Engelcke, I Posner, M Osborne; International Conference on Machine Learning (ICML), 2019
“Vote3Deep: Fast Object Detection in 3D Point Clouds Using Efficient Convolutional Neural Networks” – M Engelcke, D Rao, DZ Wang, CH Tong, I Posner; IEEE International Conference on Robotics and Automation (ICRA), 2017
Other Publications
“3D Semantic Segmentation with Submanifold Sparse Convolutional Networks” – B Graham, M Engelcke, L van der Maaten; Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2018
“Large-Scale 3D Shape Reconstruction and Segmentation from ShapeNet Core55” – L Yi, L Shao, M Savva, et al.; arXiv preprint arXiv:1710.06104, 2017
Videos
Blog Posts
DeepSets: Modeling Permutation Invariance (Guest post on inFERENCe.vc)
Scaling up Neural Networks for Processing 3D Point Clouds
Open-Source Contributions
