Views
No views yet
| file | what it is | input → output |
|---|---|---|
vda_vits_encoder_518x910_fp16.onnx (45 MB) | per-frame DINOv2-S encoder | [1,3,518,910] normalized RGB → 4 × [1,2405,384] ViT features |
vda_vits_head_T32_37x65_fp16.onnx (14 MB) | temporal DPT head, 32-frame window | 4 × [32,2405,384] stacked features → [32,1,518,910] depth |
LICENSE). The Base/Large variants are CC-BY-NC and are
not included. This repo contains a format conversion (ONNX/fp16) of the original
checkpoint; no weights were retrained or fine-tuned.1@inproceedings{video_depth_anything,
2 title={Video Depth Anything: Consistent Depth Estimation for Super-Long Videos},
3 author={Chen, Sili and Guo, Hengkai and Zhu, Shengnan and Zhang, Feihu and Huang, Zilong and Feng, Jiashi and Kang, Bingyi},
4 booktitle={CVPR},
5 year={2025}
6}