prs-eth/Marigold
Repurposes diffusion-based image generators for monocular depth estimation and other image analysis tasks including normals and intrinsics.
Not currently ranked — collecting fresh signals.
star history
Marigold adapts pretrained diffusion models, originally designed for image generation, into image analysis tools for depth estimation and related perception tasks. The method fine-tunes the denoising process to extract geometric information like depth maps, surface normals, and intrinsic image components from single RGB images. It operates in a zero-shot, in-the-wild setting and provides models via Hugging Face for inference.
Frequently asked
- What is prs-eth/Marigold?
- Repurposes diffusion-based image generators for monocular depth estimation and other image analysis tasks including normals and intrinsics.
- Is Marigold open source?
- Yes — prs-eth/Marigold is open source, released under the Apache-2.0 license.
- What language is Marigold written in?
- prs-eth/Marigold is primarily written in Python.
- How popular is Marigold?
- prs-eth/Marigold has 3.2k stars on GitHub.
- Where can I find Marigold?
- prs-eth/Marigold is on GitHub at https://github.com/prs-eth/Marigold.