prs-eth/marigold-disparity-affine-v0-1

Marigold Disparity v0.1 Model Card

The model is fine-tuned from the stable-diffusion-2 model as described in our papers, in inverse depth (disparity) space, with "trailing" and "zero-snr":

• CVPR'2024 paper titled "Repurposing Diffusion-Based Image Generators for Monocular Depth Estimation"
• Journal extension titled "Marigold: Affordable Adaptation of Diffusion-Based Image Generators for Image Analysis"

Using the model

• This model is for internal test.
• Get to the bottom of things with our official codebase.
• Developed by: Bingxin Ke, Anton Obukhov, Shengyu Huang, Nando Metzger, Rodrigo Caye Daudt, Konrad Schindler.
• Model type: Generative latent diffusion-based affine-invariant disparity (inverse depth) estimation from a single image.
• Language: English.
• License: Apache License License Version 2.0.
• Cite as:

@misc{ke2025marigold,
  title={Marigold: Affordable Adaptation of Diffusion-Based Image Generators for Image Analysis},
  author={Bingxin Ke and Kevin Qu and Tianfu Wang and Nando Metzger and Shengyu Huang and Bo Li and Anton Obukhov and Konrad Schindler},
  year={2025},
  eprint={2505.09358},
  archivePrefix={arXiv},
  primaryClass={cs.CV}
}

@InProceedings{ke2023repurposing,
  title={Repurposing Diffusion-Based Image Generators for Monocular Depth Estimation},
  author={Bingxin Ke and Anton Obukhov and Shengyu Huang and Nando Metzger and Rodrigo Caye Daudt and Konrad Schindler},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  year={2024}
}