DepthAnything/Video-Depth-Anything

Health v2 · maintenance only

Adoption not part of the score

2087 stars · 193 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

Video Depth Anything is a transformer-based monocular depth estimation model for arbitrarily long videos, built on Depth Anything V2. It provides temporally consistent relative and metric depth with faster inference and fewer parameters than diffusion-based alternatives.

Use cases

• estimate depth maps from long videos
• generate temporally consistent video depth
• recover metric depth from monocular video
• unproject video frames to point clouds
• stream depth estimation for long videos
• benchmark video depth models

When to choose

• you need consistent depth for super-long videos
• you want faster, lighter inference than diffusion-based depth models
• you need both relative and metric depth options

When to avoid

• you only need single-image depth without temporal consistency
• you need real-time depth on CPU without a GPU

Facets

library · maturity active

machine-learning computer-vision video-processing image-processing computer-vision machine-learning artificial-intelligence python depth-estimation monocular-depth video-depth transformer metric-depth streaming-inference cvpr-2025 video gpu

Member repositories

For agents

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