Computer Vision and 3D Perception
Applications of LLMs in computer vision tasks, 3D perception, scene understanding, and visual data processing for engineering applications
Computer Vision and 3D Perception
Research on Large Language Models in Computer Vision and 3D Perception
Enhancing 3D Perception with MLLMs
Unlocking spatial understanding from 2D images using advanced language models
Vocabulary-Free 3D Scene Understanding
Revolutionizing 3D Instance Segmentation with Open-Ended Reasoning
Eagle: Advancing Visual Understanding in AI
Optimizing multimodal LLMs through mixture of vision encoders
Hier-SLAM: Scaling Up 3D Semantic Mapping
Hierarchical categorical Gaussian splatting for efficient semantic SLAM
Combating Hallucinations in Video AI
A new benchmark for detecting false events in Video LLMs
AI-Powered CAD: Vision to 3D Design
Using Vision-Language Models to Generate CAD Code from Visual Inputs
Transforming Segmentation Through Language
Converting complex image segmentation into simple text generation
GraphCLIP: The Foundation Model for Text-Attributed Graphs
Enhancing cross-domain transferability in graph analysis without extensive labeling
Scaling 3D Scene Understanding
A breakthrough dataset for indoor 3D vision models
Accelerating Vision-Language Models
Reducing Visual Redundancy for Faster LVLMs
Accelerating Vision AI with LoRA Optimization
Reducing computational costs while enhancing performance of multimodal models
Multimodal CAD Generation Revolution
Unifying text, image, and point cloud inputs for CAD model creation
FlexCAD: Revolutionizing Design Generation
Unified control across all CAD construction hierarchies using fine-tuned LLMs
Intelligent Physics Simulation for 3D Environments
Using Multi-Modal LLMs to Guide Gaussian Splatting in Physics Simulations
Smart Physics for Virtual Objects
Automating Material Properties in 4D Content Generation
PerLA: The Next Leap in 3D Understanding
A perceptive AI system that captures both fine details and global context in 3D environments
Next-Level 3D Interaction
Teaching AI to understand sequential actions in 3D environments
ARCON: Next-Generation Video Prediction
Auto-Regressive Continuation for Enhanced Driving Video Generation
Olympus: The Universal Computer Vision Router
Transforming MLLMs into a unified framework for diverse visual tasks
Revolutionizing Physics Simulations
A More Stable Material Point Method for Engineering Applications
LLaVA-UHD v2: Advancing Visual Intelligence in AI
Enhancing fine-grained visual perception through hierarchical processing
ECBench: Testing AI's Understanding of Human Perspective
A new benchmark for evaluating how well AI models comprehend egocentric visual data
Accelerating Vision-Language Models
A Global Compression Strategy for High-Resolution VLMs
FALCON: Revolutionizing Visual Processing in MLLMs
Solving High-Resolution Image Challenges with Visual Registers
Vision Meets Language: Next-Gen Semantic Segmentation
How Large Language Models Enhance Visual Scene Understanding
The Future of 3D Point Cloud Models
Advancing foundation models for complex 3D environments
Visual Intelligence in Text-to-CAD Generation
Enhancing LLMs with visual feedback for precise CAD modeling
Scaling Visual Attention Across GPUs
Efficient cross-attention for processing large visual inputs in multimodal AI
Open-Vocabulary Articulated 3D Objects
Converting static meshes to articulated objects with natural language guidance
Combating AI Visual Hallucinations
Steering Visual Information to Reduce False Claims in LVLMs
AI-Powered CAD Editing Revolution
First framework for automated text-based CAD modification
Building Digital Twins: The Future of Urban Planning
Integrating 3D Modeling, GIS, and AI for Virtual Building Replicas
Revolutionizing Imaging with Hyperspectral AI
Transforming invisible data into actionable insights across industries
Event Cameras for Text Recognition
Pioneering bio-inspired solutions for challenging visual environments
AI-Powered Traffic Monitoring
Integrating Multimodal-LLMs with Instance Segmentation for Smarter Cities
Hierarchical 3D Semantic Mapping
Advanced SLAM with Hierarchical Categorical Gaussian Splatting
Teaching Robots to Understand 3D Worlds
Using LLMs to Detect Object Affordances in Open Environments
Master-Apprentice VLM Inference
Reducing costs while maintaining high-quality vision-language responses
Combating AI Hallucinations with Octopus
A dynamic approach to reduce fabricated responses in vision-language models
Accelerating Multimodal AI: DivPrune
Cutting Visual Token Count While Preserving Performance
3D Scene Intelligence
Enhancing LLM Geometric Reasoning for Object Placement
SmartWay: Revolutionizing AI Navigation
Enhanced Waypoint Prediction with Intelligent Backtracking
HybridAgent: Intelligent Image Restoration
Unifying Multiple Restoration Models for Superior Results
PiSA: Revolutionizing 3D Understanding in AI
Self-Augmenting Data to Bridge the 3D-Language Gap
Bridging 2D and 3D Vision-Language Understanding
Overcoming 3D data scarcity with unified architecture
ChatStitch: Visualizing the Blind Spots
Advancing Collaborative Vehicle Perception with LLM Agents
Smart 3D Spatial Understanding for Robots
Leveraging LLMs to Create Hierarchical Scene Graphs for Indoor Navigation
Open Vocabulary 3D Scene Understanding
Advancing 3D segmentation with CLIP and superpoints
Matrix: Voice-Driven 3D Objects in AR
Transforming spoken words into interactive 3D environments
Revolutionizing Road Infrastructure Maintenance
AI-Powered Crack Detection for Urban Digital Twins
3D Open Vocabulary Scene Understanding
Advancing Security with Instance-Level 3D Scene Analysis
Bridging 2D to 3D: MLLMs for Spatial Reasoning
Transferring 2D image understanding to 3D scene segmentation
Revolutionizing CAD Drawing Analysis
Progressive Hierarchical Tuning for Efficient Parametric Primitive Analysis
Evaluating AI Vision for Self-Driving Cars
Fine-grained assessment framework for autonomous driving AI systems
3D Vision Meets Natural Language
Advancing Autonomous Vehicles' Understanding of Human Instructions
Smart Robots for Accessible Environments
Enhancing Assistive Technologies with Open-Vocabulary Semantic Understanding
3D Visual Reasoning Breakthrough
Leveraging LLMs for Enhanced Object Detection in Complex Environments
LLM-Guided Architecture Evolution
Autonomous AI-powered optimization for object detection models
Advancing Robot Vision with AI Fusion
Leveraging Multimodal Fusion and Vision-Language Models for Smarter Robots
Smarter, Smaller Vision Transformers
Strategic Pruning for Domain Generalization and Resource Efficiency
Explaining the Uncertainty Gap in AI Vision
Revealing why image classifiers lack confidence through counterfactuals
Q-Agent: Quality-Driven Image Restoration
A robust multimodal LLM approach for handling complex image degradations
Accelerating Large Vision Language Models
Reducing computational overhead through strategic token pruning
