Accepted Papers

Why Do Self-Supervised Models Transfer? On the Impact of Invariance on Downstream Tasks

Linus Ericsson (University of Edinburgh)*et. al.

BMVC 2022 Accepted Paper / Poster

Automatic universal taxonomies for multi-domain semantic segmentation

Petra Bevandić (University of Zagreb)*et. al.

BMVC 2022 Accepted Paper / Poster

Best Paper Award

Positive Pair Distillation Considered Harmful: Continual Meta Metric Learning for Lifelong Object Re-Identification

Kai Wang (Universitat Autònoma de Barcelona)*et. al.

BMVC 2022 Accepted Paper / Poster

Why 3D Object \& Scene Representations Are Needed for Scene Understanding

Chris Williams (University of Edinburgh)

Augmenting Object Detection Supervised Training with Pixel-level Contrastive Learning

YASSER ABDELAZIZ DAHOU DJILALI (Dublin City University)*et. al.

Negative mining in self-supervised action localisation

Dezhao Luo (Queen Mary, University of London)*et. al.

Challenging universal representation of deep models against a simple baseline method for 3D point cloud registration

David Bojanić (University of Zagreb)*et. al.

Representational Structure of Neural Networks Trained on Biased and Out-Of-Distribution Data

Gnyanesh Bangaru (VNRVJIET)*et. al.