Pascal Mettes

Latest

• Accuracy of deep learning-based AI models for early caries lesion detection: the influence of annotation quality and reference choice
• Minimizing Hyperbolic Embedding Distortion with LLM-guided Hierarchy Restructuring
• Large Language Models Are Natural Video Popularity Predictors
• Low-distortion and GPU-compatible Tree Embeddings in Hyperbolic Space
• Designing Hierarchies for Optimal Hyperbolic Embedding
• Hyperbolic Safety-Aware Vision-Language Models
• Microtomography to traditional dental radiograph: Projecting 3D initial proximal caries lesion annotations for enhanced radiographic delineation
• SimZSL: Zero-Shot Learning Beyond a Pre-defined Semantic Embedding Space
• BrainACTIV: Identifying visuo-semantic properties driving cortical selectivity using diffusion-based image manipulation
• Compositional entailment learning for hyperbolic vision-language models
• Union-over-Intersections: Object Detection beyond Winner-Takes-All
• Enhancement of early proximal caries annotations in radiographs: introducing the Diagnostic Insights for Radiographic Early-caries with micro-CT (ACTA-DIRECT) dataset
• Hyp 2 Nav: Hyperbolic Planning and Curiosity for Crowd Navigation
• Hyperbolic Random Forests
• Intriguing Properties of Hyperbolic Embeddings in Vision-Language Models
• Hyperbolic Deep Learning in Computer Vision: A Survey
• Flow Matching for Conditional Text Generation in a Few Sampling Steps
• Latent Space Editing in Transformer-Based Flow Matching
• Focus for Free in Density-Based Counting
• Find the Cliffhanger: Multi-modal Trailerness in Soap Operas
• Infinite Class Mixup
• HypLL: The Hyperbolic Learning Library
• Poincaré ResNet
• Universal Prototype Transport for Zero-Shot Action Recognition and Localization
• Multi-Label Meta Weighting for Long-Tailed Dynamic Scene Graph Generation
• Are 3D convolutional networks inherently biased towards appearance?
• Hyperbolic Graph Codebooks
• Multi-Task Edge Prediction in Temporally-Dynamic Video Graphs
• Teaching a New Dog Old Tricks: Contrastive Random Walks in Videos with Unsupervised Priors
• Maximum Class Separation as Inductive Bias in One Matrix
• Less than Few: Self-Shot Video Instance Segmentation
• Hyperbolic Image Segmentation
• On Measuring and Controlling the Spectral Bias of the Deep Image Prior
• Hyperbolic Busemann Learning with Ideal Prototypes
• Diagnosing Errors in Video Relation Detectors
• Zero-Shot Action Recognition from Diverse Object-Scene Compositions
• Social Fabric: Tubelet Compositions for Video Relation Detection
• Unsharp Mask Guided Filtering
• Few-Shot Transformation of Common Actions into Time and Space
• Object Priors for Classifying and Localizing Unseen Actions
• Adversarial Self-Supervised Scene Flow Estimation
• Interactivity Proposals for Surveillance Videos
• Trajectory Prediction Network for Future Anticipation of Ships
• Heterogeneous Non-Local Fusion for Multimodal Activity Recognition
• Open Cross-Domain Visual Search
• Localizing the Common Action Among a Few Videos
• PointMixup: Augmentation for Point Clouds