LibreDepthPro2b-depth

Apple Depth Pro 2 (DINOv3 ViT-B/16) bundled with Apple's companion focal-length estimator (DepthPro2-focal) for metric single-image and streaming video depth estimation, repackaged in LibreYOLO checkpoint format.

NON-COMMERCIAL RESEARCH-ONLY WEIGHTS

These weights are not covered by LibreYOLO's MIT license.

Apple Machine Learning Research Model is licensed under the Apple Machine Learning Research Model License Agreement.

Under the Apple Machine Learning Research Model License Agreement (apple-amlr, Copyright (C) 2026 Apple Inc. All Rights Reserved), use, modification, and redistribution of these pretrained weights are limited exclusively to non-commercial Research Purposes (scientific research and academic development; no commercial exploitation, product development, or use in any commercial product or service). Downstream users are solely responsible for complying with the license terms.

Source

  • Depth checkpoint: apple/DepthPro2-B model.safetensors at pinned revision 85e67fdfb7f5b36a6f6c4495e6789d331b1d8a04 (SHA-256: 244ff85e0d9bde6e0f985de1c61420abc4c2110265b495c974556225ac2b60ed).
  • Focal-length checkpoint: apple/DepthPro2-focal model.safetensors at pinned revision 622659d44a2002cf9ae4333f2569afa938290d9a (SHA-256: 8f567d1c97ff916137288668f6bd306ce2094da2e2b852b0e4539dbb139c8c86).
  • Upstream reference code: apple-aiml-research/ml-depth-pro-2 at commit aa389cbb0b74341664ed663a0966c859905cd6c0 (Apple Sample Code License).

Modifications

None to the learned parameters: every tensor from apple/DepthPro2-B and apple/DepthPro2-focal is preserved unchanged (bit-identical inference parity verified against upstream). The only modifications are bundling the apple/DepthPro2-focal state dict under the fov.* key prefix alongside the depth state dict and adding LibreYOLO checkpoint v1.0 metadata (task="depth").

See weights/convert_depthpro2_weights.py and weights/upload_depthpro2_hf.py in the LibreYOLO source repository.

Usage

from libreyolo import LibreYOLO

model = LibreYOLO("LibreDepthPro2b-depth.pt")
result = model.predict("image.jpg")
depth = result.depth_map            # metric depth in metres (inf on sky pixels)
K = depth.intrinsics                # (3, 3) OpenCV pixel intrinsics

License

Apple Machine Learning Research Model License Agreement (apple-amlr). Apple Machine Learning Research Model is licensed under the Apple Machine Learning Research Model License Agreement. See the verbatim LICENSE and NOTICE files in this repository.

Downloads last month
22
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for LibreYOLO/LibreDepthPro2b-depth

Finetuned
(3)
this model