Skip to content
Live
Trending storyPast story

FactEmbed enhances the performance of combined image retrieval

1 reports1 outlets15 days agoUpdate

The story so far

AI overview

FactEmbed decouples query understanding from embedding construction, utilizing decomposed latent vectors to simulate visual modifications before retrieval, significantly improving performance in composite image retrieval. This method outperforms existing technologies in the CIR benchmark, optimizing decomposed representations through intermediate target imagination and leveraging global-level embeddings and mean pooling of latent vectors for efficient inference.

AI-written from the reports · Updated 14 days ago

Coverage timeline

Follow the coverage to see every side of the story.

4 Nov
  1. Hugging Face Daily Papers
    Factorized MLLM Embeddings for Composed Image Retrieval

    FactEmbed decouples query understanding from embedding construction, simulating pre-retrieval visual modifications using decomposed latent vectors, significantly improving performance in composed image retrieval. The method outperforms existing technologies in the CIR benchmark, optimizing decomposed representations through intermediate target imagination, and achieving efficient inference using global-level embeddings and mean pooling of latent vectors.