ankanbhunia/PIDM
A diffusion model that synthesizes person images from source images and target poses.
Not currently ranked — collecting fresh signals.
star history
PIDM is a denoising diffusion probabilistic model for generating person images given a source image and a target pose. It leverages a U-Net based denoising architecture combined with pose-guided feature modulation to synthesize realistic human images. The model was published at CVPR 2023 and provides both training code and a Google Colab demo for inference.
Frequently asked
- What is ankanbhunia/PIDM?
- A diffusion model that synthesizes person images from source images and target poses.
- Is PIDM open source?
- Yes — ankanbhunia/PIDM is open source, released under the MIT license.
- What language is PIDM written in?
- ankanbhunia/PIDM is primarily written in Jupyter Notebook.
- How popular is PIDM?
- ankanbhunia/PIDM has 503 stars on GitHub.
- Where can I find PIDM?
- ankanbhunia/PIDM is on GitHub at https://github.com/ankanbhunia/PIDM.