Cvpr 2026 Findings 2026

761 papers

Top Research Directions

Research directions most represented by Cvpr 2026 Findings 2026 papers.

The Decoupling of Visual Perception and Logical Reasoning 10
The Spatial Reasoning Bottleneck in Multimodal Models 9
Generalizing from Synthetic Data to the Real World 9
Scaling Visual Attention Beyond Quadratic Complexity 9
The Cross Modal Safety Gap in Multimodal Models 7
Scaling Spatiotemporal Attention Beyond Quadratic Complexity 7
Group Relative Policy Optimization for Large Reasoning Models 6
Spatial Attention Masking in Generative Diffusion Models 6
Modality Imbalance in Multimodal Large Language Models 6
Managing Visual Token Redundancy in Multimodal Models 6
The Representational Bottleneck of Frozen Foundation Models 5
Overcoming Low Frequency Bias with Spectral Losses 5
Closed Loop Visual Feedback for Multimodal Generation 5
Aligning Video Generation Models with Physical Laws 5
Zero-Shot Generalization through Diverse Synthetic Supervision 5

Top Authors

AuthorPapers
Andrew Feng1
Peter Anthony Beerel1
Zhenglong Sun1

Papers

Title
Thinking with Blueprints: Assisting Vision–Language Models in Spatial Reasoning via Structured Object Representation
Think Twice, Act Once: Verifier-Guided Action Selection For Embodied Agents
THOM: Generating Physically Plausible Hand-Object Meshes From Text
Through the PRISM: Principle-Aware, Interpretable, and Multi-Scale Evaluation of Visual Designs
TICON: A Slide-Level Tile Contextualizer for Histopathology Representation Learning
Tiny Inference-Time Scaling with Latent Verifiers
TinySR: Shallow Diffusion Transformers for Real-World Image Super-Resolution
TokenErase: Robust Concept Erasure via Visual-Injected Token Optimization
Towards Calibrated Gradient-Based Multi-Task Learning
Towards Complete Activation: Foreground-Background Multi-Perspective Guided Cross-Support for Few-Shot Segmentation
Towards Efficient Multimodal Unified Reasoning Model via Model Merging
Towards Generalization of Scene Text Tampering Localization via Causal Invariance
Towards Imperceptible Watermarking Via Environment Illumination for Consumer Cameras
Towards Metric-Aware Multi-Person Mesh Recovery by Jointly Optimizing Human Crowd in Camera Space
Towards Noise-Robust Medical Segmentation via Chebyshev-Attention-Based Asymmetric UNet
Towards Reliable Human Evaluations in Gesture Generation: Insights from a Community-Driven State-of-the-Art Benchmark
Towards Robust Content Watermarking Against Removal and Forgery Attacks
Towards Source-Aware Object Swapping with Initial Noise Perturbation
Towards Text-Guided Attribute-Disentangled Multimodal Representation Learning
Towards Universal Open-Set Visual Font Recognition Via Augmented Synthetic Similarity
TP-Seg: Task-Prototype Framework for Unified Medical Lesion Segmentation
TPTransformer: Tensor–Tensor Product Transformer for Hyperspectral Image Super-Resolution
Training-Free Cross-Modal Alignment via Anchor Profiles with Statistical Significance Testing
Training-Free Uncertainty-Guided Logit Adjustment for Few-Shot Class-Incremental Learning
Trajectory-Diversity-Driven Robust Vision-and-Language Navigation
TransKV: A Data-Driven Pruning Method for Large Foundation Models
TriGuard-FL: A User-Centric Trust Triad in Federated Learning via Auditable Data, Verifiable Contributions, and Antidote-Driven Mitigation
Turning Generators into Retrievers: Unlocking MLLMs for Natural Language-Guided Geo-Localization
UDVideoQA: A Traffic Video Question Answering Dataset for Multi-Object Spatio-Temporal Reasoning in Urban Dynamics
UGLMM: Towards Unified Vision Grounding with Large Multimodal Model
UI-AGILE: Advancing GUI Agents with Effective Reinforcement Learning and Precise Inference-Time Grounding
UMI-HOI: Unifying Multimodal Information with Semantic Multi-Head Attention for Human–Object Interaction Detection
Unbiased Dynamic Multimodal Fusion
Uncertainty-Aware Cross-Modal Opinion Interaction: A General Frameworkfor Visible-Infrared Vehicle and Person Re-Identification
Uncertainty-Guided Graph Formulation via MWIS for Token Pruning in LVLMs
Understanding Reward Hacking in Text-to-Image Reinforcement Learning
UnfoldIR: Rethinking Deep Unfolding Network in Illumination Degradation Image Restoration
UniD-Shift: Towards Unified Semantic Segmentation via Interpretable Shared–Private Multimodal Decomposition
Unified Urban Tuning: Co-Enhancing Satellite and Street View Reasoning with a Progressive Tuning Framework
UNIFORM: Unifying Knowledge from Large-Scale and Diverse Pre-Trained Models
Unifying Scientific Communication: Fine-Grained Correspondence Across Scientific Media
Unify the Views: View-Consistent Prototype Learning for Few-Shot Segmentation
UniLat3D: Geometry-Appearance Unified Latents for Single-Stage 3D Generation
UniLayDiff: A Unified Diffusion Transformer for Content-Aware Layout Generation
UniShield: An Adaptive Multi-Agent Framework for Unified Forgery Image Detection and Localization
UniTalking: A Unified Audio-Video Framework for Talking Portrait Generation
UniVerse3D: Emerging Properties of Unified Multimodal Models in 3D Understanding and Generation
Unleashing the Potential of Event-Based Stereo Via Coarse-to-Fine Bio-Inspired Regression
Unlocking ImageNet’s Multi-Object Nature: Automated Large-Scale Multilabel Annotation
Unlocking Single-View Constraints for Efficient Camera Relocalization with Keypoint-Level Multi-View Geometric Consistency in Training

« prev Page 14/16 next »