| Video-Depth-Anything-Base -- unofficial transformers-format package (trust_remote_code) |
| Packaged by bigshanedogg, 2026. NOT an official release of the original authors. |
|
|
| Components |
| ---------- |
| 1. Video Depth Anything Base weights (video_depth_anything_vitb.pth) |
| Source: https://huggingface.co/depth-anything/Video-Depth-Anything-Base |
| Copyright: Sili Chen, Hengkai Guo, Shengnan Zhu, Feihu Zhang, Zilong Huang, Jiashi Feng, Bingyi Kang (ByteDance) |
| License: CC-BY-NC-4.0 (non-commercial) -- LICENSE |
| 2. Video Depth Anything code (video_depth_anything/ incl. its DINOv2 copy, utils/util.py; vendored into vda_*.py) |
| Source: https://github.com/DepthAnything/Video-Depth-Anything @ 4f5ae23172ba60fd7bc11ef671cca678842c7072 |
| Copyright: Copyright (2025) Bytedance Ltd. and/or its affiliates; DINOv2 layers Copyright (c) Meta Platforms, Inc. and affiliates. |
| License: Apache-2.0 -- LICENSE.code |
|
|
| Changes made in this package |
| ---------------------------- |
| - vda_*.py: the official files concatenated into flat files; package-internal imports replaced by relative imports; NestedTensorBlock spelled out where dinov2.py imported it as Block. No computation changed. |
| - configuration_ / modeling_ / image_processing_ / processing_video_depth_anything.py: new transformers wrappers; the model runs the official VideoDepthAnything forward, infer_video_depth and (streaming) infer_video_depth_one; the image processor is infer_video_depth's per-frame transform. |
| - model.safetensors: the .pth state dict converted by save_pretrained, names unchanged, every tensor checked equal. |
|
|
| License files |
| ------------- |
| - LICENSE: Creative Commons Attribution-NonCommercial 4.0 International (weights) |
| - LICENSE.code: Apache License 2.0, copy from DepthAnything/Video-Depth-Anything (code) |
|
|
