Visual Words Meet BM25: Sparse Auto-Encoder Visual Word Scoring for Image Retrieval

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Han, Donghoon, Park, Eunhwan, Seo, Seunghyeon
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866911490427060224
author Han, Donghoon
Park, Eunhwan
Seo, Seunghyeon
author_facet Han, Donghoon
Park, Eunhwan
Seo, Seunghyeon
contents Dense image retrieval is accurate but offers limited interpretability and attribution, and it can be compute-intensive at scale. We present \textbf{BM25-V}, which applies Okapi BM25 scoring to sparse visual-word activations from a Sparse Auto-Encoder (SAE) on Vision Transformer patch features. Across a large gallery, visual-word document frequencies are highly imbalanced and follow a Zipfian-like distribution, making BM25's inverse document frequency (IDF) weighting well suited for suppressing ubiquitous, low-information words and emphasizing rare, discriminative ones. BM25-V retrieves high-recall candidates via sparse inverted-index operations and serves as an efficient first-stage retriever for dense reranking. Across seven benchmarks, BM25-V achieves Recall@200 $\geq$ 0.993, enabling a two-stage pipeline that reranks only $K{=}200$ candidates per query and recovers near-dense accuracy within $0.2$\% on average. An SAE trained once on ImageNet-1K transfers zero-shot to seven fine-grained benchmarks without fine-tuning, and BM25-V retrieval decisions are attributable to specific visual words with quantified IDF contributions.
format Preprint
id arxiv_https___arxiv_org_abs_2603_05781
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Visual Words Meet BM25: Sparse Auto-Encoder Visual Word Scoring for Image Retrieval
Han, Donghoon
Park, Eunhwan
Seo, Seunghyeon
Computer Vision and Pattern Recognition
Artificial Intelligence
Dense image retrieval is accurate but offers limited interpretability and attribution, and it can be compute-intensive at scale. We present \textbf{BM25-V}, which applies Okapi BM25 scoring to sparse visual-word activations from a Sparse Auto-Encoder (SAE) on Vision Transformer patch features. Across a large gallery, visual-word document frequencies are highly imbalanced and follow a Zipfian-like distribution, making BM25's inverse document frequency (IDF) weighting well suited for suppressing ubiquitous, low-information words and emphasizing rare, discriminative ones. BM25-V retrieves high-recall candidates via sparse inverted-index operations and serves as an efficient first-stage retriever for dense reranking. Across seven benchmarks, BM25-V achieves Recall@200 $\geq$ 0.993, enabling a two-stage pipeline that reranks only $K{=}200$ candidates per query and recovers near-dense accuracy within $0.2$\% on average. An SAE trained once on ImageNet-1K transfers zero-shot to seven fine-grained benchmarks without fine-tuning, and BM25-V retrieval decisions are attributable to specific visual words with quantified IDF contributions.
title Visual Words Meet BM25: Sparse Auto-Encoder Visual Word Scoring for Image Retrieval
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2603.05781