# DepthAnything/PromptDA

[CVPR 2025] Prompt Depth Anything

Repository: https://github.com/DepthAnything/PromptDA
Canonical: https://ross.abutalabs.com/products/promptda
Homepage: https://promptda.github.io
Language: Python
License: Apache-2.0
License Family: permissive
Topics: 3d-reconstruction, 4d-reconstruction, depth-estimation, robotics-grasping
Last push: 2026-01-29T16:53:35+00:00

## Health v2 (maintenance only)
Score: 50/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 64, release rhythm 35, longevity 44
- inputs: {"age_days": 624, "days_push": 216, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1159, forks 73 (observed 2026-08-28T04:03:48.588715+00:00)

## What it is
Prompt Depth Anything is a Python library implementing a CVPR 2025 method for high-resolution (up to 4K) accurate metric depth estimation. It uses low-cost LiDAR depth (e.g., iPhone ARKit) as a prompt to guide a Depth Anything foundation model, with pretrained checkpoints and a Hugging Face demo.

## Use cases
- estimate accurate metric depth from a single image using iPhone LiDAR as a prompt
- generate 4K resolution depth maps for 3D reconstruction
- get depth priors for robotic grasping
- improve depth estimation on transparent objects
- benchmark monocular depth estimation on ARKitScenes and ScanNet++
- upscale low-resolution LiDAR depth to full image resolution

## When to choose
- you need metric (meter-scale) depth, not just relative depth
- you have access to a low-resolution LiDAR depth prompt from an iPhone or similar sensor
- you want high-resolution depth maps for 3D reconstruction or robot grasping pipelines
- you want a research-grade model with pretrained checkpoints and a paper backing it

## When to avoid
- you have no LiDAR or other prompt depth source available
- you need a lightweight production model for CPU-only or mobile deployment
- you only need relative/monocular depth without metric scale
- you need a stable long-term-supported API rather than a research codebase

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, computer-vision, image-processing
- domain: computer-vision, deep-learning, robotics, artificial-intelligence
- platform: python
- tags: depth-estimation, metric-depth, lidar, monocular-depth, 3d-reconstruction, robotic-grasping, cvpr-2025, depth-anything, 4k-resolution, pretrained-models, gpu, linux, macos

## Member repositories
- DepthAnything/PromptDA (main) score 50

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:48.588715+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T06:31:51.084853+00:00, confidence not recorded.
  - readme: https://github.com/DepthAnything/PromptDA (fetched 2026-08-28T04:03:48.588715+00:00, sha 9e55981e5766)
  - homepage: https://promptda.github.io (fetched 2026-08-29T12:36:31.514724+00:00, sha 49e7108800e7)
- Data as of 2026-08-30T08:39:29.467469+00:00.