WildDet3D

WildDet3D predicts three-dimensional boxes from a single image.

Tasks: 3d detection
Sizes: l at 1008 px
Install: pip install "libreyolo[hf]"
Support tier: Sibling tier, since v1.6.0. A separate product surface with its own factory and contract.
Upstream: WildDet3D by Allen Institute for AI, SAM License. Paper, source
Licenses: Code MIT, weights SAM License. Commercial use

Install

bash

pip install "libreyolo[hf]"

Predict

Python

from libreyolo import LibreWildDet3D, SAMPLE_IMAGEfrom PIL import Imageimport numpy as np # Requires the separately installed upstream runtime.model = LibreWildDet3D(device="cpu")# Approximate pinhole intrinsics for the original image size.# Replace with your camera's measured 3x3 calibration for real geometry.width, height = Image.open(SAMPLE_IMAGE).sizef = max(width, height)intrinsics = np.array([[f, 0, width / 2], [0, f, height / 2], [0, 0, 1]], dtype=np.float32)result = model.predict(SAMPLE_IMAGE, intrinsics=intrinsics, text="person")print(result.boxes3d)

WildDet3D accepts text, box or point prompts and requires original-image camera intrinsics. Install its separate upstream runtime and provide runtime_path and runtime_python when they are outside the active environment.

result.boxes3d holds centers, dimensions and wxyz quaternions in camera coordinates, plus confidence, class and intrinsics. Plotting projects cuboids into the image. Training, validation, tracking and export are not supported. See 3D detection.

Licensing

Check the license on the Hugging Face repository of the specific weights you download. Every checkpoint in the LibreYOLO org carries one, and they are not always the same across a family. That repository is the authoritative source; the summary below describes what applied when this page was last verified.

This is a description of the licenses involved, not legal advice. If the answer matters commercially, read the licenses yourself and take your own counsel.

Original work: WildDet3D, Allen Institute for AI
Upstream license: SAM License
Upstream source: github.com/allenai/WildDet3D
LibreYOLO code: MIT
Weights: SAM License, republished at huggingface.co/LibreYOLO
Interpretation: The upstream runtime and checkpoint retain the custom SAM License, including its field-of-use restrictions.

Citation

@misc{huang2026wilddet3dscalingpromptable3d,
      title={WildDet3D: Scaling Promptable 3D Detection in the Wild}, 
      author={Weikai Huang and Jieyu Zhang and Sijun Li and Taoyang Jia and Jiafei Duan and Yunqian Cheng and Jaemin Cho and Matthew Wallingford and Rustin Soraki and Chris Dongjoo Kim and Shuo Liu and Donovan Clay and Taira Anderson and Winson Han and Ali Farhadi and Bharath Hariharan and Zhongzheng Ren and Ranjay Krishna},
      year={2026},
      eprint={2604.08626},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2604.08626}, 
}

Copied from the authors' citation block at raw.githubusercontent.com/allenai/WildDet3D/main/README.md.