Clothing agnostic Pre-inpainting Virtual Try-ON
📖Paper -- 📖Paper arxiv -- 💾Code -- 🕹️Colab_Demo
With the development of deep learning technology, virtual try-on technology has developed important application value in the fields of e-commerce, fashion, and entertainment. The recently proposed Leffa technology has addressed the texture distortion problem of diffusion-based models, but there are limitations in that the bottom detection inaccuracy and the existing clothing silhouette persist in the synthesis results. To solve this problem, this study proposes CaP-VTON (Clothing-Agnostic Pre-Inpainting Virtual Try-On). CaP-VTON integrates DressCode-based multi-category masking and Stable Diffusion-based skin inflation preprocessing; in particular, a generated skin module was introduced to solve skin restoration problems that occur when long-sleeved images are converted to short-sleeved or sleeveless ones, introducing a preprocessing structure that improves the naturalness and consistency of full-body clothing synthesis and allowing the implementation of high-quality restoration considering human posture and color. As a result, CaP-VTON achieved 92.5%, which is 15.4% better than Leffa, in short-sleeved synthesis accuracy and consistently reproduced the style and shape of the reference clothing in visual evaluation. These structures maintain model-agnostic properties and are applicable to various diffusion-based virtual inspection systems; they can also contribute to applications that require high-precision virtual wearing, such as e-commerce, custom styling, and avatar creation.
Create a Conda Python environment and install requirements.
It runs on Linux (Ubuntu) environment...!
conda create -n capvton python==3.10
conda activate capvton
cd CAP-VTON
pip install -r requirements.txt
"Run Start_CaP_VTON.ipynb"
- Capacity: 34.3GB
- RAM: 24GB more
- GPU: (Tested: RTX4080 and A100)
This work was developed by extending Leffa.
We would like to acknowledge the contributions of the original authors.
For in-depth technical details, please see the Leffa Paper.
If you find CAP-VTON helpful for your research, please cite our work:
@article{DevChoco_CAP-VTON_2025,
author = {Sehyun, Kim. Hye Jun, Lee. Jiwoo, Lee. Taemin, Lee.},
title = {Clothing-Agnostic Pre-Inpainting Virtual Try-On},
journal = {Electronics},
year = {2025},
volume = {14},
number = {23},
article-number= {4710},
pages = {4710},
doi = {10.3390/electronics14234710},
url = {https://www.mdpi.com/2079-9292/14/23/4710},
publisher = {MDPI}
}