Anna Kukleva
YOU?
Author Swipe
View article: RefAM: Attention Magnets for Zero-Shot Referral Segmentation
RefAM: Attention Magnets for Zero-Shot Referral Segmentation Open
Most existing approaches to referring segmentation achieve strong performance only through fine-tuning or by composing multiple pre-trained models, often at the cost of additional training and architectural modifications. Meanwhile, large-…
View article: Language-Unlocked ViT (LUViT): Empowering Self-Supervised Vision Transformers with LLMs
Language-Unlocked ViT (LUViT): Empowering Self-Supervised Vision Transformers with LLMs Open
The integration of Large Language Model (LLMs) blocks with Vision Transformers (ViTs) holds immense promise for vision-only tasks by leveraging the rich semantic knowledge and reasoning capabilities of LLMs. However, a fundamental challeng…
T-FAKE: Synthesizing Thermal Images for Facial Landmarking Open
Facial analysis is a key component in a wide range of applications such as healthcare, autonomous driving, and entertainment. Despite the availability of various facial RGB datasets, the thermal modality, which plays a crucial role in life…
View article: X-MIC: Cross-Modal Instance Conditioning for Egocentric Action Generalization
X-MIC: Cross-Modal Instance Conditioning for Egocentric Action Generalization Open
Lately, there has been growing interest in adapting vision-language models (VLMs) to image and third-person video classification due to their success in zero-shot recognition. However, the adaptation of these models to egocentric videos ha…
View article: OrCo: Towards Better Generalization via Orthogonality and Contrast for Few-Shot Class-Incremental Learning
OrCo: Towards Better Generalization via Orthogonality and Contrast for Few-Shot Class-Incremental Learning Open
Few-Shot Class-Incremental Learning (FSCIL) introduces a paradigm in which the problem space expands with limited data. FSCIL methods inherently face the challenge of catastrophic forgetting as data arrives incrementally, making models sus…
In-Style: Bridging Text and Uncurated Videos with Style Transfer for Text-Video Retrieval Open
Large-scale noisy web image-text datasets have been proven to be efficient for learning robust vision-language models. However, when transferring them to the task of video retrieval, models still need to be fine-tuned on hand-curated paire…
Temperature Schedules for Self-Supervised Contrastive Methods on Long-Tail Data Open
Most approaches for self-supervised learning (SSL) are optimised on curated balanced datasets, e.g. ImageNet, despite the fact that natural data usually exhibits long-tail distributions. In this paper, we analyse the behaviour of one of th…
TAEC: Unsupervised Action Segmentation with Temporal-Aware Embedding and Clustering Open
Temporal action segmentation in untrimmed videos has gained increased attention recently. However, annotating action classes and frame-wise boundaries is extremely time consuming and cost intensive, especially on large-scale datasets. To a…
Learning by Sorting: Self-supervised Learning with Group Ordering Constraints Open
Contrastive learning has become an important tool in learning representations from unlabeled data mainly relying on the idea of minimizing distance between positive data pairs, e.g., views from the same images, and maximizing distance betw…
Leveraging Self-Supervised Training for Unintentional Action Recognition Open
Unintentional actions are rare occurrences that are difficult to define precisely and that are highly dependent on the temporal context of the action. In this work, we explore such actions and seek to identify the points in videos where th…
Chronic biopsy proven post‐COVID myoendocarditis with SARS‐Cov‐2 persistence and high level of antiheart antibodies Open
Purpose To study the clinical signs and mechanisms (viral and autoimmune) of myoendocarditis in the long‐term period after COronaVIrus Disease 2019 (COVID‐19). Methods Fourteen patients (nine male, 50.1 ± 10.2 y.o.) with biopsy proven post…
Thromboembolic mask of severe lymphocytic myopericarditis: possibilities of clinical and morphological diagnostics and complex treatment Open
This report presents a clinical case of a 57-year-old female patient who was admitted for dyspnea, productive cough, reduced left ventricular (LV) systolic function, and who had previously undergone thoracocentesis for significant pleural …
Infective and nonbacterial thrombotic endocarditis in patients with post-COVID-19 viral-immune myocarditis Open
The possibility of heart inflammation (both myocardial and endocardial) months after a coronavirus disease 2019 (COVID-19) has not been practically studied, especially since approaches to the treatment of myocarditis in combination with va…
View article: CycDA: Unsupervised Cycle Domain Adaptation from Image to Video
CycDA: Unsupervised Cycle Domain Adaptation from Image to Video Open
Although action recognition has achieved impressive results over recent years, both collection and annotation of video training data are still time-consuming and cost intensive. Therefore, image-to-video adaptation has been proposed to exp…
Post-COVID morphologically proved endocarditis: infective and nonbacterial forms Open
Purpose To study clinical features of endocarditis and its possible mechanisms (infective and nonbacterial) in the long-term period after acute COVID-19. Methods Three patients (two male and one female, age 64, 39 and 46 years) diagnosed w…
Post-COVID myocarditis diagnosed by endomyocardial biopsy and/or magnetic resonance imaging 2–9 months after acute COVID-19 Open
Purpose To study clinical features of myocarditis and its possible mechanisms (including persistence of SARS-Cov-2 in the myocardium) in the long-term period after acute COVID-19. Methods Fifteen patients (8 male and 7 female, mean age 47.…
Progressive chronic SARS‐CoV‐2‐positive giant cell myoendocarditis with atrial standstill and sudden cardiac death Open
Giant cell myocarditis (GCM) is a rare condition. Its association with SARS‐CoV‐2 has not been described before. The 46‐year‐old female patient was admitted to the clinic on September 2020. She had 7 year adrenal insufficiency history and …
Subacute and chronic post-covid myoendocarditis: clinical presentation, role of coronavirus persistence and autoimmune mechanisms Open
Aim To study clinical features of myoendocarditis and its possible mechanisms, including persistence of SARS-Cov-2 in the myocardium, in the long-term period following COVID-19. Material and methods This cohort, prospective study included …
Revisiting Consistency Regularization for Semi-supervised Learning Open
Consistency regularization is one of the most widely-used techniques for semi-supervised learning (SSL). Generally, the aim is to train a model that is invariant to various data augmentations. In this paper, we revisit this idea and find t…
PATHOLOGICAL ANATOMY OF INFECTION CAUSED BY SARS-COV-2 Open
Autopsy data from 80 patients who died of the COVID-19 infection were analysed. Using macro- and microscopic studies, specific features of pathological processes in various organs were identified. The obtained experimental data, along with…
Miocarditis in Patients with COVID-19 Confirmed by Immunohistochemical Open
Aim Despite the regular heart damage in patients with coronavirus pneumonia caused by SARS-Cov-2, a possibility of developing lymphocytic myocarditis as a part of COVID-19 remains unsubstantiated. The aim of this study was to demonstrate a…
Joint Visual-Temporal Embedding for Unsupervised Learning of Actions in\n Untrimmed Sequences Open
Understanding the structure of complex activities in untrimmed videos is a\nchallenging task in the area of action recognition. One problem here is that\nthis task usually requires a large amount of hand-annotated minute- or even\nhour-lon…