Mateusz Koziński
YOU?
Author Swipe
View article: Harnessing deep learning to detect bronchiolitis obliterans syndrome from chest CT
Harnessing deep learning to detect bronchiolitis obliterans syndrome from chest CT Open
Our approach shows potential for improving BOS diagnosis by enabling early detection and management. The ability to detect BOS from standard-resolution scans at any stage of respiration makes this method more accessible than previous appro…
View article: Partial annotations in active learning for semantic segmentation
Partial annotations in active learning for semantic segmentation Open
View article: MULDE: Multiscale Log-Density Estimation via Denoising Score Matching for Video Anomaly Detection
MULDE: Multiscale Log-Density Estimation via Denoising Score Matching for Video Anomaly Detection Open
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. This lets us estimate …
View article: Harnessing Deep Learning to Detect Bronchiolitis Obliterans Syndrome from Chest CT
Harnessing Deep Learning to Detect Bronchiolitis Obliterans Syndrome from Chest CT Open
Bronchiolitis Obliterans Syndrome (BOS), a fibrotic airway disease following lung transplantation, conventionally relies on pulmonary function tests (PFTs) for diagnosis due to limitations of CT images. Thus far, deep neural networks (DNNs…
View article: Sit Back and Relax: Learning to Drive Incrementally in All Weather Conditions
Sit Back and Relax: Learning to Drive Incrementally in All Weather Conditions Open
In autonomous driving scenarios, current object detection models show strong performance when tested in clear weather. However, their performance deteriorates significantly when tested in degrading weather conditions. In addition, even whe…
View article: LaFTer: Label-Free Tuning of Zero-shot Classifier using Language and Unlabeled Image Collections
LaFTer: Label-Free Tuning of Zero-shot Classifier using Language and Unlabeled Image Collections Open
Recently, large-scale pre-trained Vision and Language (VL) models have set a new state-of-the-art (SOTA) in zero-shot visual classification enabling open-vocabulary recognition of potentially unlimited set of categories defined as simple l…
View article: MAtch, eXpand and Improve: Unsupervised Finetuning for Zero-Shot Action Recognition with Language Knowledge
MAtch, eXpand and Improve: Unsupervised Finetuning for Zero-Shot Action Recognition with Language Knowledge Open
Large scale Vision-Language (VL) models have shown tremendous success in aligning representations between visual and text modalities. This enables remarkable progress in zero-shot recognition, image generation & editing, and many other exc…
View article: Persistent Homology With Improved Locality Information for More Effective Delineation
Persistent Homology With Improved Locality Information for More Effective Delineation Open
Persistent Homology (PH) has been successfully used to train networks to detect curvilinear structures and to improve the topological quality of their results. However, existing methods are very global and ignore the location of topologica…
View article: Video Test-Time Adaptation for Action Recognition
Video Test-Time Adaptation for Action Recognition Open
Although action recognition systems can achieve top performance when evaluated on in-distribution test points, they are vulnerable to unanticipated distribution shifts in test data. However, test-time adaptation of video action recognition…
View article: ActMAD: Activation Matching to Align Distributions for Test-Time-Training
ActMAD: Activation Matching to Align Distributions for Test-Time-Training Open
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. We propose to perform this adaptation via Activation Matching (ActMAD): We analy…
View article: MATE: Masked Autoencoders are Online 3D Test-Time Learners
MATE: Masked Autoencoders are Online 3D Test-Time Learners Open
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. Like existing TTT methods from the 2D image …
View article: Adjusting the Ground Truth Annotations for Connectivity-Based Learning to Delineate
Adjusting the Ground Truth Annotations for Connectivity-Based Learning to Delineate Open
Deep learning-based approaches to delineating 3D structure depend on accurate annotations to train the networks. Yet, in practice, people, no matter how conscientious, have trouble precisely delineating in 3D and on a large scale, in part …
View article: Enforcing connectivity of 3D linear structures using their 2D projections
Enforcing connectivity of 3D linear structures using their 2D projections Open
Many biological and medical tasks require the delineation of 3D curvilinear structures such as blood vessels and neurites from image volumes. This is typically done using neural networks trained by minimizing voxel-wise loss functions that…
View article: TOPO-Loss for continuity-preserving crack detection using deep learning
TOPO-Loss for continuity-preserving crack detection using deep learning Open
We present a method for segmenting cracks in images of masonry buildings damaged by earthquakes. Existing methods of crack detection fail to preserve the continuity of cracks, and their performance deteriorates with imprecise training labe…
View article: Dataset for TOPO-Loss for continuity-preserving crack detection using deep learning
Dataset for TOPO-Loss for continuity-preserving crack detection using deep learning Open
This is the dataset used to assess the performance of the crack detection algorithm proposed by Pantoja-Rosero et, al (2022) in the article "TOPO-Loss for continuity-preserving crack detection using deep learning" (https://doi.org/10.1016/…
View article: Dataset for TOPO-Loss for continuity-preserving crack detection using deep learning
Dataset for TOPO-Loss for continuity-preserving crack detection using deep learning Open
This is the dataset used to assess the performance of the crack detection algorithm proposed by Pantoja-Rosero et, al (2022) in the article "TOPO-Loss for continuity-preserving crack detection using deep learning" (https://doi.org/10.1016/…
View article: Generating LOD3 building models from structure-from-motion and semantic segmentation
Generating LOD3 building models from structure-from-motion and semantic segmentation Open
This paper describes a pipeline for automatically generating level of detail (LOD) models (digital twins), specifically LOD2 and LOD3, from free-standing buildings. Our approach combines structure from motion (SfM) with deep-learning-based…
View article: Dataset for generating LOD3 building models from structure-from-motion and semantic segmentation
Dataset for generating LOD3 building models from structure-from-motion and semantic segmentation Open
This repository contains the codes for computing geometrical digital twins as LOD3 models for buildings, using a structure from motion and semantic segmentation. The methodology hereby implements was presented in the paper [Generating LOD3…
View article: Dataset for generating LOD3 building models from structure-from-motion and semantic segmentation
Dataset for generating LOD3 building models from structure-from-motion and semantic segmentation Open
This repository contains the codes for computing geometrical digital twins as LOD3 models for buildings, using a structure from motion and semantic segmentation. The methodology hereby implements was presented in the paper [Generating LOD3…
View article: Enforcing Connectivity of 3D Linear Structures Using Their 2D Projections
Enforcing Connectivity of 3D Linear Structures Using Their 2D Projections Open
View article: Training set of microscopy images for Dietler et al. Nature Communications 2020
Training set of microscopy images for Dietler et al. Nature Communications 2020 Open
Training set of microscopy images for Dietler et al. Nature Communications 2020
View article: Training set of microscopy images for Dietler et al. Nature Communications 2020
Training set of microscopy images for Dietler et al. Nature Communications 2020 Open
Training set of microscopy images for Dietler et al. Nature Communications 2020
View article: Adjusting the Ground Truth Annotations for Connectivity-Based Learning\n to Delineate
Adjusting the Ground Truth Annotations for Connectivity-Based Learning\n to Delineate Open
Deep learning-based approaches to delineating 3D structure depend on accurate\nannotations to train the networks. Yet, in practice, people, no matter how\nconscientious, have trouble precisely delineating in 3D and on a large scale,\nin pa…
View article: Localized Persistent Homologies for more Effective Deep Learning.
Localized Persistent Homologies for more Effective Deep Learning. Open
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. However, existing methods are very global and ig…
View article: Persistent Homology with Improved Locality Information for more Effective Delineation
Persistent Homology with Improved Locality Information for more Effective Delineation Open
Persistent Homology (PH) has been successfully used to train networks to detect curvilinear structures and to improve the topological quality of their results. However, existing methods are very global and ignore the location of topologica…
View article: Drainage Canals in Southeast Asian Peatlands Increase Carbon Emissions
Drainage Canals in Southeast Asian Peatlands Increase Carbon Emissions Open
Drainage canals associated with logging and agriculture dry out organic soils in tropical peatlands, thereby threatening the viability of long‐term carbon stores due to increased emissions from decomposition, fire, and fluvial transport. I…
View article: Drainage Canals in Southeast Asian Peatlands Increase Carbon Emissions
Drainage Canals in Southeast Asian Peatlands Increase Carbon Emissions Open
Earth and Space Science Open Archive This preprint has been submitted to and is under consideration at AGU Advances. ESSOAr is a venue for early communication or feedback before peer review. Data may be preliminary.Learn more about preprin…
View article: Promoting Connectivity of Network-Like Structures by Enforcing Region Separation
Promoting Connectivity of Network-Like Structures by Enforcing Region Separation Open
We propose a novel, connectivity-oriented loss function for training deep convolutional networks to reconstruct network-like structures, like roads and irrigation canals, from aerial images. The main idea behind our loss is to express the …
View article: A convolutional neural network segments yeast microscopy images with high accuracy
A convolutional neural network segments yeast microscopy images with high accuracy Open
The identification of cell borders (‘segmentation’) in microscopy images constitutes a bottleneck for large-scale experiments. For the model organism Saccharomyces cerevisiae , current segmentation methods face challenges when cells bud, c…
View article: Promoting Connectivity of Network-Like Structures by Enforcing Region\n Separation
Promoting Connectivity of Network-Like Structures by Enforcing Region\n Separation Open
We propose a novel, connectivity-oriented loss function for training deep\nconvolutional networks to reconstruct network-like structures, like roads and\nirrigation canals, from aerial images. The main idea behind our loss is to\nexpress t…