/VPD

[ICCV 2023] VPD is a framework that leverages the high-level and low-level knowledge of a pre-trained text-to-image diffusion model to downstream visual perception tasks.

Primary LanguageJupyter NotebookMIT LicenseMIT

Stargazers