Aligning Forest and Trees in Images & Long Captions for Visually Grounded Understanding
cs.CV, cs.AI, cs.LG
Submitted: 2026-02-03
Updated: 2026-11-10
Terminology
Sources
- An Exploration of Hierarchical Attention Transformers for Efficient Long Document Classification
- HiVLP: Hierarchical Vision-Language Pre-Training for Fast Image-Text Retrieval
- An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
- ImageInWords: Unlocking Hyper-Detailed Image Descriptions
- HiCLIP: Contrastive Language-Image Pretraining with Hierarchy-aware Attention
- HDT: Hierarchical Document Transformer
- PowerCLIP: Powerset Alignment for Contrastive Pre-Training
- CLIP Is Shortsighted: Paying Attention Beyond the First Sentence
- Modeling Caption Diversity in Contrastive Vision-Language Pretraining
- Superpixel Tokenization for Vision Transformers: Preserving Semantic Integrity in Visual Tokens
- Decoupled Weight Decay Regularization
- TULIP: Token-length Upgraded CLIP
- Visually Consistent Hierarchical Image Classification
- Hierarchical Context Merging: Better Long Context Understanding for Pre-trained LLMs
- EVA-CLIP: Improved Training Techniques for CLIP at Scale
- SigLIP 2: Multilingual Vision-Language Encoders with Improved Semantic Understanding, Localization, and Dense Features
- Hi-Transformer: Hierarchical Interactive Transformer for Efficient and Effective Long Document Modeling
- PixCLIP: Achieving Fine-grained Visual Language Understanding via Any-granularity Pixel-Text Alignment Learning
- FG-CLIP: Fine-Grained Visual and Textual Alignment
- Demystifying CLIP Data
Related papers
- TEA: Text Encoder Alignment for Robust Concept Erasure in Text-to-Image Models
- Do Egocentric Video-Language Models Capture Both Hand- and Object-Centric Cues?
- Aura: Consistent Multi-Subject Video Generation via VLM-Grounded Semantic Alignment
- PVRF: All-in-one Adverse Weather Removal via Prior-modulated and Velocity-constrained Rectified Flow
- DLWM: Diverse Latent World Models for Efficient Multimodal Reasoning
- Out of Sight, Still in Mind: Token Compression for Omni-LLMs