Thomas Verelst

Deep Learning Engineer @ Axelera AI

PhD (KU Leuven, Belgium). Efficient neural networks with dynamic CNNs and conditional computation.
Enabling deep learning and computer vision on edge devices.

Resume/CV | Google Scholar

Want to know more about conditional neural networks? Check out my reading list!

First-Author Publications

BlockCopy: High-Resolution Video Processing with Block-Sparse Feature Propagation and Online Policies

Thomas Verelst, Tinne Tuytelaars

ICCV 2021 - main conference poster

[paper] [code]
SegBlocks: Block-Based Dynamic Resolution Networks for Real-Time Segmentation

Thomas Verelst, Tinne Tuytelaars

ECCV 2020 - Embedded Vision Workshop

Best Short Paper Award received on August 28th 2020

Extended version under review

[ECCV short paper] [IEEE long paper]
Dynamic Convolutions: Exploiting Spatial Sparsity for Faster Inference

Thomas Verelst, Tinne Tuytelaars

CVPR 2020

[paper] [Pytorch code] [YouTube 1-minute teaser]
Generating superpixels with deep representations

Thomas Verelst, Matthew Blaschko, Maxim Berman

CVPR 2018 workshop on DeepVision: Beyond supervised learning (extended abstract)
[paper]

Other work

Processor Architecture Optimization for Spatially Dynamic Neural Networks

Steven Colleman, Thomas Verelst, Linyan Mei, Tinne Tuytelaars, Marian Verhelst

2021 IFIP/IEEE 29th International Conference on Very Large Scale Integration (VLSI-SoC)
[paper]
“Can human behaviour, emotion and cognition be detected by computer vision techniques? A use case on student engagement”

Pieter Vanneste, José Oramas, Thomas Verelst, Tinne Tuytelaars, Annelies Raes, Fien Depaepe, Wim Van den Noortgate

Mathematics 2021 journal paper
[paper]