Machine Learning & Computer Vision

Publications, preprints & participation to conferences

• Spatial Consistency Loss for Training Multi-Label Classifiers from Single-Label Annotations, Thomas Verelst, Paul K. Rubenstein, Marcin Eichner, Tinne Tuytelaars, Maxim Berman. arXiv preprint, March 2022.
• PhD Thesis: Tractable Approximations for Achieving Higher Model Efficiency in Computer Vision Defended in September 2020
• AOWS: Adaptive and optimal network width search with latency constraints, Maxim Berman, Leonid Pishchulin, Ning Xu, Matthew B. Blaschko, Gérard Medioni, CVPR 2020 (oral)
  • Paper on CVF
  • Oral video
• Discriminative training of conditional random fields with probably submodular constraints, Maxim Berman, Matthew B. Blaschko. IJCV, January 2020. [preprint]
• Function Norms for Neural Networks, Amal Rannen Triki, Maxim Berman, Vladimir Kolmogorov, Matthew B. Blaschko. ICCV 2019 workshop on Statistical Deep Learning for Computer Vision
• A Bayesian Optimization Framework for Neural Network Compression, Xingchen Ma, Amal Rannen Triki, Maxim Berman, Christos Sagonas, Jacques Cali, Matthew B. Blaschko. ICCV 2019
• Optimizing the Dice Score and Jaccard Index for Medical Image Segmentation: Theory & Practice, Jeroen Bertels, Tom Elbode, Maxim Berman, Dirk Vandermeulen, Frederik Maes, Raf Bisschops, Matthew B. Blaschko. MICCAI 2019
• Adaptive Compression-based Lifelong Learning, Shivangi Srivastava, Maxim Berman, Matthew B. Blaschko, Devis Tuia, BMVC 2019 spotlight
• MultiGrain: a unified image embedding for classes and instances, [code], Maxim Berman, Hervé Jégou, Andrea Vedaldi, Iasonas Kokkinos, Matthijs Douze. arXiv preprint. Work done during an internship at Facebook AI Research in Paris.
• The Lovász-Softmax loss: A tractable surrogate for the optimization of the intersection-over-union measure in neural networks, Maxim Berman, Amal Rannen Triki, Matthew B. Blaschko. Published in CVPR 2018.
  • Project page
  • Code
• Masters thesis I supervised at KU Leuven (2018)
  • Generating superpixels with deep representations (extended abstract), Thomas Verelst, Maxim Berman, Matthew B. Blaschko, poster at "Deep Vision" workshop at CVPR 2018.
  • Efficient semantic image segmentation with superpixel pooling, Mathijs Schuurmans, Maxim Berman, Matthew B. Blaschko, arxiv preprint. [Code]
• Stochastic Weighted Function Norm Regularization, Amal Rannen Triki, Maxim Berman, Matthew B. Blaschko, arxiv preprint, Oct. 2017
• Efficient optimization for probably submodular constraints in CRFs, Maxim Berman, Matthew B. Blaschko. Presented at the NIPS workshop on Constructive Machine Learning 2016.
• Monocular Surface Reconstruction using 3D Deformable Part Models, Stefan Kinauer, Maxim Berman, Iasonas Kokkinos, presented at "Geometry Meets Deep Learning" in ECCV 2016.

My Master works

• Master Internship report: higher-order grammars in Computer Vision
• Paper study on Local Coordinate Coding

Physics

Master works

• Internship report: Cold Atoms stirring

Bachelor works

• Internship report: Simulation de la dynamique moléculaire des disques durs
• TIPE sur l'holographie

On GitHub

• Pytorch-DAG simple DAG module in Pytorch
• Enet-PyTorch Convert a lua model into PyTorch, compatible with ENet
• SSVMjulia.jl -- Implementation of Structured SVMs in Julia
• Daisy.jl -- C wrapper around C++ Daisy Library, and Julia integration

Misc

I based the Pelican theme for this website on pelican-bootstrap3; modifications can be made available.