About

I work as a Research Scientist at Facebook AI Research London. Previously I was a DPhil student at the Visual Geometry Group, University of Oxford funded by Naver Labs Europe. My supervisors were Diane Larlus (Naver Labs Europe) and Andrea Vedaldi (University of Oxford).

My research focuses on discovering and representing the semantic and geometric structure of object categories from images or videos.

Publications

Reizenstein, J. and Shapovalov, R. and Henzler, P. and Sbordone, L. and Labatut, P. and Novotny, D.
Common Objects in 3D: Large-Scale Learning and Evaluation of Real-life 3D Category Reconstruction
ICCV 2021, oral presentation, Best Paper Honorable Mention
  pdf   code   dataset   project page   talk

Henzler, P. and Reizenstein, J. and Labatut, P. and Shapovalov, R. and Ritschel, T. and Vedaldi, A. and Novotny, D.
Unsupervised Learning of 3D Object Categories from Videos in the Wild
CVPR 2021
  pdf   cvf   poster   talk

Eisenberger, M. and Novotny, D. and Kerchenbaum, G. and Labatut, P. and Neverova, N. and Cremers, D. and Vedaldi, A.
NeuroMorph: Unsupervised Shape Interpolation and Correspondence in One Go
CVPR 2021
  pdf   cvf

Biggs, B. and Ehrhadt, S. and Joo, H. and Graham, B. and Vedaldi, A. and Novotny, D.
3D Multi-bodies: Fitting Sets of Plausible 3D Human Models to Ambiguous Image Data
NeurIPS 2020, spotlight presentation
  pdf   project page   NeurIPS   talk

animated

Graham, B. and Novotny, D.
RidgeSfM: Structure from Motion via Robust Pairwise Matching Under Depth Uncertainty
3DV 2020
  pdf   code

Novotny, D.* and Shapovalov, R.* and Vedaldi, A.
Canonical 3D Deformer Maps: Unifying Parametric and Non-parametric Methods for Dense Weakly-supervised Category Reconstruction
NeurIPS 2020
  pdf   poster   code   NeurIPS

Neverova, N. and Novotny, D. and Khalidov, V. and Szafraniec, M. and Labatut, P. and Vedaldi, A.
Continuous Surface Embeddings
NeurIPS 2020
  pdf   NeurIPS

Novotny, D. and Graham, B. and Reizenstein, J.
PerspectiveNet: A Scene-consistent Image Generator for New View Synthesis in Real Indoor Environments
NeurIPS 2019
  pdf   poster   NeurIPS

animated

Novotny, D.* and Ravi, N.* and Graham, B. and Neverova, N. and Vedaldi, A.
C3DPO: Canonical 3D Pose Networks for Non-Rigid Structure From Motion
ICCV 2019, oral presentation
  pdf   code   poster   talk   cvf

animated

Novotny, D. and Larlus, D. and Vedaldi, A.
Capturing the Geometry of Object Categories from Video Supervision
TPAMI 2018
  pdf   ieee

Novotny, D.* and Albanie, S.* and Larlus, D. and Vedaldi, A.
Semi-convolutional Operators for Instance Segmentation
ECCV 2018
  pdf   poster   cvf

Novotny, D.* and Albanie, S.* and Larlus, D. and Vedaldi, A.
Self-supervised Learning of Geometrically Stable Features Through Probabilistic Introspection
CVPR 2018, spotlight
  pdf   poster   talk (starts at 1:25:00)   cvf

Novotny, D. and Larlus, D. and Vedaldi, A.
Learning 3D Object Categories by Looking Around Them
ICCV 2017, oral presentation
  pdf   poster   talk   scholar

Novotny, D. and Larlus, D. and Vedaldi, A.
AnchorNet: A Weakly Supervised Network to Learn Geometry-sensitive Features For Semantic Matching
CVPR 2017
  pdf   poster   code   scholar

Novotny, D. and Larlus, D. and Vedaldi, A.
I Have Seen Enough: Transferring Parts Across Categories
BMVC 2016
  pdf   poster   scholar   dataset

Novotny, D. and Larlus, D. and Vedaldi, A.
Learning the Structure of Objects from Web Supervision
ECCV 2016 Workshops, Oral presentation
  pdf   poster   scholar

Novotny, D. and Matas, J.
Cascaded Sparse Spatial Bins for Efficient and Effective Generic Object Detection
ICCV 2015
  pdf   poster   video   scholar

Other

Novotny, D. and Larlus, D. and Perronnin, F. and Vedaldi, A.
Understanding the Fisher Vector: a Multimodal Part Model
arXiv 2015
  pdf   scholar