We present X-MDPT (Cross-view Masked Diffusion Prediction Transformers), a novel diffusion model designed for pose-guided human image generation. X-MDPT distinguishes itself by employing masked diffusion transformers that operate on latent patches, a departure from the commonly-used Unet structures in existing works. The model comprises three key modules: 1) a denoising diffusion Transformer, 2) an aggregation network that consolidates conditions into a single vector for the diffusion process, and 3) a mask cross-prediction module that enhances representation learning with semantic information from the reference image. X-MDPT demonstrates scalability, improving FID, SSIM, and LPIPS with larger models. Despite its simple design, our model outperforms state-of-the-art approaches on the DeepFashion dataset while exhibiting efficiency in terms of training parameters, training time, and inference speed. Our compact 33MB model achieves an FID of 7.42, surpassing a prior Unet latent diffusion approach (FID 8.07) using only 11× fewer parameters. Our best model surpasses the pixel-based diffusion with 32 of the parameters and achieves 5.43× faster inference. The code is available at https://github.com/trungpx/xmdpt.
@article{arxiv.2402.01516,
title = {Cross-view Masked Diffusion Transformers for Person Image Synthesis},
author = {Trung X. Pham and Zhang Kang and Chang D. Yoo},
journal= {arXiv preprint arXiv:2402.01516},
year = {2024}
}