Video Depth Anything Small, streaming ONNX export (154x266)

Streaming ONNX export of Video Depth Anything Small, used by Spatial-Canvas for temporally stable in-browser video depth.

Upstream code: https://github.com/DepthAnything/Video-Depth-Anything (commit 4f5ae23172ba60fd7bc11ef671cca678842c7072).
Weights: depth-anything/Video-Depth-Anything-Small, video_depth_anything_vits.pth (Apache-2.0; checkpoint sha256 in export-manifest.json). Only the Small model is Apache-2.0; Base/Large are CC-BY-NC and are NOT used here.
Modification: re-exported to ONNX at a fixed 154x266 input with a 31-frame temporal cache split into encoder, bootstrap and stream graphs, converted to fp16. Export details, shapes and parity checks: export-manifest.json. No retraining.
License: Apache-2.0, see LICENSE.

Downloads last month: -; Downloads are not tracked for this model. How to track

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support