| --- |
| license: apache-2.0 |
| library_name: libreyolo |
| tags: |
| - depth-estimation |
| - depth-anything |
| --- |
| |
| # LibreDepthAnythingV2s-depth |
|
|
| Depth Anything V2 Small (ViT-S encoder + DPT head), repackaged for LibreYOLO's `depth` task. |
|
|
| ## Source |
|
|
| Derived from [DepthAnything/Depth-Anything-V2](https://github.com/DepthAnything/Depth-Anything-V2) |
| at commit `03876f8651c73a60fe4c2c48294e09fcb6838fcf`. |
| Copyright (c) 2024 Depth Anything V2 authors (TikTok / ByteDance). Licensed under the Apache License, Version 2.0. |
|
|
| The ViT-S encoder is a DINOv2-Small backbone from [facebookresearch/dinov2](https://github.com/facebookresearch/dinov2). |
| Copyright (c) Meta Platforms, Inc. and affiliates. Licensed under the Apache License, Version 2.0. |
|
|
| ## Modifications |
|
|
| State-dict key remapping only. Learned parameters are unchanged. |
| See `weights/convert_depth_anything_v2_weights.py` in the |
| [LibreYOLO source repository](https://github.com/LibreYOLO/libreyolo). |
|
|
| ## Usage |
|
|
| ```python |
| from libreyolo import LibreYOLO |
| |
| model = LibreYOLO("LibreYOLO/LibreDepthAnythingV2s-depth") |
| results = model.predict("image.jpg") |
| depth_map = results[0].depth_map # DepthMap(shape=(H, W), higher = closer) |
| ``` |
|
|
| > **Note — weight licensing split:** The Apache-2.0 license covers the Depth Anything V2 |
| > *source code* and the *Small (ViT-S) checkpoint* only. Base / Large / Giant checkpoints |
| > are CC-BY-NC-4.0 (non-commercial). LibreYOLO does not redistribute those; convert them |
| > locally with `weights/convert_depth_anything_v2_weights.py`. |
|
|
| ## License |
|
|
| Apache License 2.0. See the [`LICENSE`](./LICENSE) and [`NOTICE`](./NOTICE) files in this repository. |
|
|
