Skip to content
Other

LibreDepthPro2l-depth

Apple depth estimation for LibreYOLO

2 min readupdated

The model combines Apple Depth Pro 2 (built on the DINOv3 ViT-L/16 architecture) and the auxiliary DepthPro2-focal focal length estimator into a single LibreYOLO format file. The tool calculates metric depth from a single image or in streaming video, and also determines camera intrinsic parameters. The original Apple weights remain unchanged, guaranteeing results that match the source completely. This solution will be useful for computer vision and 3D reconstruction researchers.

Key features

  • Combines the apple/DepthPro2-L and apple/DepthPro2-focal checkpoints under a unified LibreYOLO interface.
  • Outputs a metric depth map in meters and a matrix of camera pixel intrinsics (optical parameters for OpenCV).
  • Weights are saved without fine-tuning, delivering bit-exact output relative to Apple's official code.
  • The Apple AMLR license permits non-commercial scientific research and development exclusively.

What you can do

  • Build a metric depth map for a single photo in an academic project.
  • Run real-time depth estimation to analyze streaming video.
  • Obtain an OpenCV camera parameter matrix (3×3) without prior manual calibration.

Example

from libreyolo import LibreYOLO

model = LibreYOLO("LibreDepthPro2l-depth.pt")
result = model.predict("image.jpg")
depth = result.depth_map            # метрическая глубина в метрах
K = depth.intrinsics                # (3, 3) интринсики камеры OpenCV
Read moreDetails
Architecture: DINOv3 ViT-L/16 + focal length estimation module (keys under the fov.* prefix).
Output details: returns infinity (inf) for sky pixels.
License: Apple Machine Learning Research Model License Agreement (non-commercial use only).
Discuss Telegram 0 21

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 1.3 GB · ready files2.0 GB
  • FP8 / INT8recommendedweights 0.7 GB1.2 GB
  • 4-bitweights 0.4 GB0.9 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: FP8 / INT8 — from RTX 2060 · BF16 — from RTX 2060

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # LibreYOLO/LibreDepthPro2l-depth
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download LibreYOLO/LibreDepthPro2l-depth --local-dir ~/models/LibreDepthPro2l-depth
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to LibreDepthPro2l-depth
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

Model family

New posts about versions of DepthPro2-L (quants, finetunes, LoRAs) will show up under the bell.

Fine-tunes 2All on Hugging Face

About the model

LibreYOLO/LibreDepthPro2l-depth
Author
LibreYOLO
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

210

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments