|
Download README.md from aether-models/aether-kit-depth-anything-3-small: direct link, hf CLI and curl.
- Browser
- Download file 973 Bytes
-
https://huggingface.co/aether-models/aether-kit-depth-anything-3-small/resolve/main/README.md
- Command line
-
hf download hf://aether-models/aether-kit-depth-anything-3-small/README.md
-
curl -L -o README.md https://huggingface.co/aether-models/aether-kit-depth-anything-3-small/resolve/main/README.md
973 Bytes
metadata
license: apache-2.0
base_model: depth-anything/DA3-SMALL
pipeline_tag: depth-estimation
tags:
- aether-kit
- core-ai
- depth-estimation
- depth-anything-3
Depth Anything 3 small for Aether-Kit (Core AI)
depth-anything/DA3-SMALL at revision e08cab65, exported by
Aether-Kit's Forge exporter (Forge/depth/) as one static Core
AI graph: an RGB image stretched to 504 x 504 on [0, 1] (ImageNet normalization is in the graph) to relative
depth (larger is farther) and a confidence map, 504 x 504 each. The DINOv2 ViT-S backbone runs in float16 and
the DualDPT head in float32, as upstream's own inference does. Load it with Aether-Kit's AetherVision
(depth-estimator@1, macOS); the manifest is .aether/aether.json. Gated against
the upstream implementation on CPU in float32 (see export-report.json).
Apache-2.0 (LICENSE). A modified version (see NOTICE).