From Pixels to Purchase: Building and Evaluating a Taxonomy-Decoupled Visual Search Engine for Home Goods E-commerce

TL;DR

This paper introduces a flexible, taxonomy-decoupled visual search engine for home goods e-commerce that improves retrieval quality and customer engagement by using classification-free proposals and an LLM-based evaluation framework.

Contribution

It presents a novel taxonomy-decoupled architecture and an LLM-as-a-Judge evaluation method, enhancing robustness and scalability in visual search systems.

Findings

Improved retrieval quality and customer engagement in a real-world deployment.

Strong correlation between offline metrics and actual business outcomes.

Effective zero-shot evaluation of visual similarity and category relevance.

Abstract

Visual search is critical for e-commerce, especially in style-driven domains where user intent is subjective and open-ended. Existing industrial systems typically couple object detection with taxonomy-based classification and rely on catalog data for evaluation, which is prone to noise that limits robustness and scalability. We propose a taxonomy-decoupled architecture that uses classification-free region proposals and unified embeddings for similarity retrieval, enabling a more flexible and generalizable visual search. To overcome the evaluation bottleneck, we propose an LLM-as-a-Judge framework that assesses nuanced visual similarity and category relevance for query-result pairs in a zero-shot manner, removing dependence on human annotations or noise-prone catalog data. Deployed at scale on a global home goods platform, our system improves retrieval quality and yields a measurable uplift in customer engagement, while our offline evaluation metrics strongly correlate with real-world outcomes.