Factorized MLLM Embeddings for Composed Image Retrieval
cs.CV, cs.IR
Submitted: 2026-11-03
Updated: 2026-11-04
Project page: https://cvlab-kaist.github.io/FactEmbed/Abstract
Terminology
Sources
- E5-V: Universal Embeddings with Multimodal Large Language Models
- GME: Improving Universal Multimodal Retrieval by Multimodal LLMs
- CompoDiff: Versatile Composed Image Retrieval With Latent Diffusion
- VLM2Vec-V2: Advancing Multimodal Embedding for Videos, Images, and Visual Documents
- VLM2Vec: Training Vision-Language Models for Massive Multimodal Embedding Tasks
- Gemini Embedding: Generalizable Embeddings from Gemini
- EmbeddingGemma: Powerful and Lightweight Text Representations
- TRACE: Task-Adaptive Reasoning and Representation Learning for Universal Multimodal Retrieval
- PLUME: Latent Reasoning Based Universal Multimodal Embedding
- Bottleneck Tokens for Unified Multimodal Retrieval
- Bridging the Gap to Real-World Object-Centric Learning
- DoRA: Weight-Decomposed Low-Rank Adaptation
- Qwen2.5-VL Technical Report
- CORE: Improving Compositional Reasoning in MLLM Embedding via Reranker Distillation
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