| --- |
| license: cc-by-nc-4.0 |
| tags: |
| - depth-anything-3 |
| - onnx |
| - depth-estimation |
| - spatial-computing |
| --- |
| |
| # Depth Anything 3 (DA3) ONNX Weights |
|
|
| This repository contains pre-exported and optimized ONNX weights for the **Depth Anything 3** model family, formatted for zero-dependency inference in [`spatialhub`](https://github.com/pankajkaushik12/spatialhub). |
|
|
| ## ⚖️ Licensing Notice |
|
|
| This repository contains models under two distinct licensing tiers. **Please adhere to the upstream licenses specified below:** |
|
|
| ### 1. Permissive / Commercial Use (Apache 2.0) |
| The following ONNX models inherit the **Apache License 2.0** from the official Depth Anything 3 release and may be used for commercial applications: |
| - `da3_small.onnx` |
| - `da3_base.onnx` |
| - `da3mono_large.onnx` |
| - `da3metric_large.onnx` |
|
|
| ### 2. Non-Commercial Use Only (CC-BY-NC 4.0) |
| The Giant variants utilize larger foundation backbones and are strictly restricted to **non-commercial, research, and educational purposes**: |
| - `da3_giant.onnx` |
| - `da3_large.onnx` |
| - `da3_nested_giant_large` (Pipeline using `da3_giant.onnx` + `da3metric_large.onnx`) |
|
|
| For commercial licensing of the Giant models, please contact the original Depth Anything 3 maintainers. |
