YvanYin/Metric3D
A monocular depth estimation model that predicts metric 3D geometry from a single image using a deep learning foundation model.
Not currently ranked — collecting fresh signals.
star history
Metric3D is an official PyTorch implementation of two research papers on monocular depth estimation and 3D reconstruction. The model uses a deep learning architecture, specifically a geometric foundation model, to predict depth maps and surface normals from single images in a zero-shot transfer setting. It supports zero-shot evaluation across different datasets without requiring domain-specific fine-tuning.
Frequently asked
- What is YvanYin/Metric3D?
- A monocular depth estimation model that predicts metric 3D geometry from a single image using a deep learning foundation model.
- Is Metric3D open source?
- Yes — YvanYin/Metric3D is open source, released under the BSD-2-Clause license.
- What language is Metric3D written in?
- YvanYin/Metric3D is primarily written in Python.
- How popular is Metric3D?
- YvanYin/Metric3D has 2.2k stars on GitHub.
- Where can I find Metric3D?
- YvanYin/Metric3D is on GitHub at https://github.com/YvanYin/Metric3D.