OpenVTON-Bench: A Large-Scale High-Resolution Benchmark for Controllable Virtual Try-On Evaluation
arXiv:2601.22725v4 Announce Type: replace-cross 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 hierar
Advances in diffusion models have significantly improved virtual try-on systems, creating a need for more robust evaluation benchmarks capable of assessing high-fidelity outputs.
Improved and standardized evaluation in VTON paves the way for more reliable and commercially viable applications, reducing development cycles and increasing adoption in retail and fashion.
The availability of a large-scale, high-resolution benchmark like OpenVTON-Bench allows for more rigorous testing and comparison of VTON models, accelerating their development and deployment.
- · Fashion Retailers
- · E-commerce
- · AI/Computer Vision Developers
- · Consumers
- · Companies with outdated VTON technologies
- · Traditional retail models (without VTON)
More accurate and realistic virtual try-on experiences become widely available for consumers.
Increased consumer confidence in online apparel purchases due to accurate virtual representation, potentially boosting e-commerce sales and reducing returns.
The development of VTON systems could integrate with other AI fashion tools, leading to highly personalized clothing design and production on demand.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.AI