DepthAnything/Depth-Anything-V2

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

8709 stars · 898 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

Depth Anything V2 is a foundation model for monocular depth estimation, trained on 595K synthetic labeled images and 62M+ real unlabeled images. It provides relative and metric depth models in multiple sizes, with faster inference and finer detail than V1 and diffusion-based alternatives.

Use cases

• estimate depth from a single image
• generate depth maps from photos
• run monocular depth estimation in python
• metric depth estimation for indoor and outdoor scenes
• fine-tune a depth estimation model on custom data
• fast depth estimation for video frames

When to choose

• you need accurate, fast monocular depth estimation from single images
• you want lightweight models (24.8M-335M params) instead of slow diffusion-based depth models
• you need both relative and metric depth models at multiple scales
• you want Hugging Face Transformers or Core ML integration

When to avoid

• you need temporally consistent depth for long videos (use Video Depth Anything instead)
• you need 4K metric depth with LiDAR prompting (use Prompt Depth Anything)
• you need stereo or multi-view depth estimation
• you cannot run GPU inference

Facets

library · maturity stable

machine-learning deep-learning computer-vision image-processing transformers computer-vision machine-learning deep-learning artificial-intelligence python cross-platform monocular-depth-estimation depth-estimation foundation-model metric-depth pretrained-models neurips-2024 gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="DepthAnything/Depth-Anything-V2")