Enterprise-ready security: ISO 27001 + SOC 2 Type I compliant.

Link to this sectionImageNet (Pseudo-Labeled) Depth Dataset#

The ImageNet (pseudo-labeled depth) dataset reuses the 1,281,167 ImageNet-1K training images, which have no ground-truth depth, and pairs each image with a pseudo depth map produced by a Depth Anything V3 teacher model (monocular and metric). It is used purely for knowledge distillation and for the scene and object diversity that ImageNet provides. As the single largest source in the roughly 2.19M-image YOLO26-Depth pretraining mix (about 58%), adding it was decisive for indoor cross-domain generalization.

Link to this sectionKey Features#

  • 1,281,167 ImageNet-1K training images covering a very broad range of objects, scenes, and contexts.
  • No ground-truth depth: every depth target is a pseudo label generated by a Depth Anything V3 teacher (monocular, metric).
  • Used purely for knowledge distillation and scene/object diversity, not as a benchmark.
  • The single largest source in the ~2.19M-image pretraining mix (~58%), decisive for indoor cross-domain generalization.
  • No validation split — validation is performed only on real ground-truth datasets.

Link to this sectionDataset Structure#

The pseudo-labeled ImageNet source consists of the ImageNet-1K training images with one machine-generated depth map per image:

  1. Training images: 1,281,167 ImageNet-1K training images, the largest single component (~58%) of the ~2.19M-image YOLO26-Depth pretraining mix.
  2. Validation: none. This source has no validation split; YOLO26-Depth validation uses only real ground-truth datasets such as NYU Depth V2 and KITTI Eigen.

Link to this sectionHow the pseudo-labels are generated#

Because ImageNet ships without depth annotations, depth targets are produced offline rather than measured:

  1. Each ImageNet training image is run through a pretrained Depth Anything V3 teacher model to predict a per-pixel metric depth map.
  2. The teacher prediction is saved as a .npy float32 array in meters, following the Ultralytics depth dataset format, and paired with its source image by file stem.
  3. The pseudo-labeled pairs are then added as one component of a larger training mix, where a student YOLO26-Depth model learns to match the teacher while also training on real ground-truth sources.

This pseudo-labeled source is a component of the internal mixed-training configuration used to pretrain the released YOLO26-Depth weights and is not distributed as a standalone download.

Link to this sectionRole in YOLO26-Depth#

This source contributes the bulk of the YOLO26-Depth pretraining data and the widest diversity of scenes and objects. By distilling a strong Depth Anything V3 teacher across more than a million images, it transfers broad scene priors into the student. In practice, adding it was decisive for indoor cross-domain generalization, helping the model perform well on indoor scenes beyond the real ground-truth training distributions.

Link to this sectionUsage#

The pseudo-labeled ImageNet data is not trained on in isolation; it is consumed as one component of a combined depth mix, defined by an internal/experiment YAML rather than a public dataset download. Conceptually, training on such a mix looks like the following:

Train Example
from ultralytics import YOLO

# Load a pretrained depth model
model = YOLO("yolo26n-depth.pt")

# Train on a combined depth mix (your own multi-source dataset YAML)
results = model.train(data="depth-mix.yaml", epochs=100, imgsz=640)

Link to this sectionPretrained Models#

Pretrained YOLO26-Depth models auto-download from the Ultralytics v8.4.0 assets release when first referenced by name:

Link to this sectionCitations and Acknowledgments#

If you use the ImageNet dataset in your research or development work, please cite the following paper:

Quote
@inproceedings{deng2009imagenet,
      title={ImageNet: A Large-Scale Hierarchical Image Database},
      author={Deng, Jia and Dong, Wei and Socher, Richard and Li, Li-Jia and Li, Kai and Fei-Fei, Li},
      booktitle={2009 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
      year={2009},
      organization={IEEE}
}

@inproceedings{yang2024depthanything,
      title={Depth Anything: Unleashing the Power of Large-Scale Unlabeled Data},
      author={Yang, Lihe and Kang, Bingyi and Huang, Zilong and Xu, Xiaogang and Feng, Jiashi and Zhao, Hengshuang},
      booktitle={IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
      year={2024}
}

We would like to acknowledge the ImageNet team for creating and maintaining the dataset, and the authors of Depth Anything for the teacher model used to generate the pseudo depth labels.

Contributors
BOBovey08091

Comments