OpenVTON-Bench: A Large-Scale High-Resolution Benchmark for Controllable Virtual Try-On Evaluation

TL;DR

OpenVTON-Bench is a comprehensive high-resolution benchmark dataset and evaluation protocol for virtual try-on systems, addressing the need for reliable, fine-grained assessment of visual quality and semantic consistency.

Contribution

The paper introduces a large-scale, high-resolution dataset with a novel multi-modal evaluation protocol combining semantic reasoning and multi-scale metrics for virtual try-on evaluation.

Findings

Strong correlation with human judgment (Kendall's τ of 0.833)

Dataset covers 20 garment categories with 100K image pairs

Proposed metrics effectively distinguish boundary errors from texture artifacts

Abstract

Recent advances in diffusion models have significantly elevated the visual fidelity of Virtual Try-On (VTON) systems, yet reliable evaluation remains a persistent bottleneck. Traditional metrics struggle to quantify fine-grained texture details and semantic consistency, while existing datasets fail to meet commercial standards in scale and diversity. We present OpenVTON-Bench, a large-scale benchmark comprising approximately 100K high-resolution image pairs (up to $1536 \times 1536$). The dataset is constructed using DINOv3-based hierarchical clustering for semantically balanced sampling and Gemini-powered dense captioning, ensuring a uniform distribution across 20 fine-grained garment categories. To support reliable evaluation, we propose a multi-modal protocol that measures VTON quality along five interpretable dimensions: background consistency, identity fidelity, texture fidelity, shape plausibility, and overall realism. The protocol integrates VLM-based semantic reasoning with a novel Multi-Scale Representation Metric based on SAM3 segmentation and morphological erosion, enabling the separation of boundary alignment errors from internal texture artifacts. Experimental results show strong agreement with human judgments (Kendall's $\tau$ of 0.833 vs. 0.611 for SSIM), establishing a robust benchmark for VTON evaluation.