IPDM-FSPS
What it is
This is a PyTorch implementation of the IPDM model for identity-preserving face sketch and photo synthesis. It uses a lightweight parallel CNN (LPCNN) and an Identity Direction Preserver (IDP) to map between face domains in latent space, followed by a diffusion process to generate the final images.
The problem it solves
Existing methods struggle to preserve identity information accurately when synthesizing faces across the large modality gap between sketches and photographs.
What you could use it for
- Generating high-quality face sketches from photos while preserving identity
- Generating realistic face photos from sketches for forensic or artistic applications
- Cross-domain face synthesis research
Read the source ↗Back to the library →
Summary written by an AI model from the source; check the source before you rely on it.