Search Results for author: Mateusz Kozinski

Found 13 papers, 8 papers with code

MULDE: Multiscale Log-Density Estimation via Denoising Score Matching for Video Anomaly Detection

1 code implementation CVPR 2024 Jakub Micorek, Horst Possegger, Dominik Narnhofer, Horst Bischof, Mateusz Kozinski

We propose a novel approach to video anomaly detection: we treat feature vectors extracted from videos as realizations of a random variable with a fixed distribution and model this distribution with a neural network.

Anomaly Detection In Surveillance Videos Denoising +2

Meta-Prompting for Automating Zero-shot Visual Recognition with LLMs

1 code implementation18 Mar 2024 M. Jehanzeb Mirza, Leonid Karlinsky, Wei Lin, Sivan Doveh, Jakub Micorek, Mateusz Kozinski, Hilde Kuehne, Horst Possegger

Prompt ensembling of Large Language Model (LLM) generated category-specific prompts has emerged as an effective method to enhance zero-shot recognition ability of Vision-Language Models (VLMs).

Language Modelling Large Language Model +1

Sit Back and Relax: Learning to Drive Incrementally in All Weather Conditions

1 code implementation30 May 2023 Stefan Leitner, M. Jehanzeb Mirza, Wei Lin, Jakub Micorek, Marc Masana, Mateusz Kozinski, Horst Possegger, Horst Bischof

We propose to store these affine parameters as a memory bank for each weather condition and plug-in their weather-specific parameters during driving (i. e. test time) when the respective weather conditions are encountered.

Autonomous Driving Incremental Learning +2

Video Test-Time Adaptation for Action Recognition

1 code implementation CVPR 2023 Wei Lin, Muhammad Jehanzeb Mirza, Mateusz Kozinski, Horst Possegger, Hilde Kuehne, Horst Bischof

Our proposed method demonstrates a substantial performance gain over existing test-time adaptation approaches in both evaluations of a single distribution shift and the challenging case of random distribution shifts.

Action Recognition Temporal Action Localization +1

ActMAD: Activation Matching to Align Distributions for Test-Time-Training

1 code implementation CVPR 2023 Muhammad Jehanzeb Mirza, Pol Jané Soneira, Wei Lin, Mateusz Kozinski, Horst Possegger, Horst Bischof

Test-Time-Training (TTT) is an approach to cope with out-of-distribution (OOD) data by adapting a trained model to distribution shifts occurring at test-time.

Image Classification

MATE: Masked Autoencoders are Online 3D Test-Time Learners

1 code implementation ICCV 2023 M. Jehanzeb Mirza, Inkyu Shin, Wei Lin, Andreas Schriebl, Kunyang Sun, Jaesung Choe, Horst Possegger, Mateusz Kozinski, In So Kweon, Kun-Jin Yoon, Horst Bischof

Our MATE is the first Test-Time-Training (TTT) method designed for 3D data, which makes deep networks trained for point cloud classification robust to distribution shifts occurring in test data.

3D Object Classification Point Cloud Classification

Enforcing connectivity of 3D linear structures using their 2D projections

1 code implementation14 Jul 2022 Doruk Oner, Hussein Osman, Mateusz Kozinski, Pascal Fua

Many biological and medical tasks require the delineation of 3D curvilinear structures such as blood vessels and neurites from image volumes.

Localized Persistent Homologies for more Effective Deep Learning

no code implementations29 Sep 2021 Doruk Oner, Adélie Garin, Mateusz Kozinski, Kathryn Hess, Pascal Fua

Persistent Homologies have been successfully used to increase the performance of deep networks trained to detect curvilinear structures and to improve the topological quality of the results.

Deep Learning

TopoAL: An Adversarial Learning Approach for Topology-Aware Road Segmentation

no code implementations ECCV 2020 Subeesh Vasu, Mateusz Kozinski, Leonardo Citraro, Pascal Fua

Instead, we use a more sophisticated discriminator that returns a label pyramid describing what portions of the road network are correct at several different scales.

Road Segmentation

Beyond the Pixel-Wise Loss for Topology-Aware Delineation

no code implementations CVPR 2018 Agata Mosinska, Pablo Marquez-Neila, Mateusz Kozinski, Pascal Fua

We propose a new loss term that is aware of the higher-order topological features of linear structures.

Cannot find the paper you are looking for? You can Submit a new open access paper.