KITTI Depth Dataset

The KITTI dataset is a real-world outdoor autonomous-driving benchmark captured from a moving vehicle in and around the city of Karlsruhe. For monocular depth estimation, the ground-truth depth is derived from a Velodyne HDL-64 LiDAR scanner and densified using the method of Uhrig et al. 2017. The resulting depth maps remain sparse, with roughly 16–20% of pixels carrying a valid depth value. KITTI is the only real outdoor long-range source in the YOLO26-Depth pretraining mix and also serves as the KITTI Eigen evaluation benchmark.

Key Features

• Real outdoor driving scenes with depths spanning up to roughly 80 m, far beyond the typical indoor range.
• Depth ground truth obtained from a Velodyne HDL-64 LiDAR and densified with the Sparsity Invariant CNNs approach of Uhrig et al. 2017.
• Sparse supervision: only about 16–20% of pixels per image carry a valid depth value; invalid pixels are masked out of the loss and metrics.
• Stereo image pairs (left image_02 and right image_03) provide additional viewpoints for training.
• Depth values are stored as .npy float32 arrays in meters, following the Ultralytics depth dataset format.

Dataset Structure

The KITTI depth data used by Ultralytics is split into two subsets:

1. Training split: 55,198 images (left image_02 and right image_03). All 28 KITTI Eigen test drives are excluded from training to keep evaluation fair.
2. Evaluation split: the KITTI Eigen test split — its 652 left-camera frames that have improved ground truth. Evaluation uses an 80 m depth cap and median (scale-only) alignment between predictions and ground truth.

The depth range reaches approximately 80 m, and the dataset YAML (depth-kitti.yaml) sets max_depth: 80 accordingly.

Role in YOLO26-Depth

KITTI supplies the only real outdoor, long-range supervision in the YOLO26-Depth pretraining mix, complementing the predominantly indoor sources. It is also the standard KITTI Eigen benchmark for reporting driving-scene depth accuracy.

KITTI is a key example of why the depth head is unbounded (log mode): a fixed 10 m output ceiling cannot represent 80 m driving scenes. See the depth task page for details on the head output range and max_depth handling.

Results

KITTI Eigen delta1 accuracy by model size (higher is better):

Measured on the 652-frame canonical split with imgsz=768 and rect=False. They are not directly comparable to published KITTI numbers: val stretches each image to a square imgsz and nearest-resamples the sparse ground truth onto it, where the reference evaluators instead resize the prediction back to the native ground-truth resolution; DepthMetrics masks ground truth against max_depth where the improved-ground-truth protocol masks gt > 0 and caps only the prediction; it pools every valid pixel of the split rather than averaging the per-image metric; and the released weights were trained with the previous split, which placed 72 of these test frames in the training set.

Dataset YAML

A YAML (Yet Another Markup Language) file is used to define the dataset configuration. It contains information about the dataset's paths, classes, and other relevant information such as the maximum depth.

!!! example "ultralytics/cfg/datasets/depth-kitti.yaml"

```yaml
--8<-- "ultralytics/cfg/datasets/depth-kitti.yaml"
```

Usage

To train a YOLO26n-Depth model on the KITTI dataset for 100 epochs, you can use the following code snippets. For a comprehensive list of available arguments, refer to the model Training page.

!!! example "Train Example"

=== "Python"

    ```python
    from ultralytics import YOLO

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

    # Train the model on KITTI
    results = model.train(data="depth-kitti.yaml", epochs=100, imgsz=640)
    ```

=== "CLI"

    ```bash
    # Start training from a pretrained *.pt model
    yolo depth train data=depth-kitti.yaml model=yolo26n-depth.pt epochs=100 imgsz=640
    ```

Pretrained Models

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

• YOLO26n-Depth
• YOLO26s-Depth
• YOLO26m-Depth
• YOLO26l-Depth
• YOLO26x-Depth

Citations and Acknowledgments

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

!!! quote ""

=== "BibTeX"

    ```bibtex
    @article{geiger2013vision,
          title={Vision meets Robotics: The KITTI Dataset},
          author={Geiger, Andreas and Lenz, Philip and Stiller, Christoph and Urtasun, Raquel},
          journal={The International Journal of Robotics Research},
          year={2013},
          publisher={SAGE Publications}
    }

    @inproceedings{uhrig2017sparsity,
          title={Sparsity Invariant CNNs},
          author={Uhrig, Jonas and Schneider, Nick and Schneider, Lukas and Franke, Uwe and Brox, Thomas and Geiger, Andreas},
          booktitle={International Conference on 3D Vision (3DV)},
          year={2017}
    }
    ```

We would like to acknowledge the Karlsruhe Institute of Technology and Toyota Technological Institute at Chicago for creating and maintaining the KITTI dataset, and Uhrig et al. for the depth densification method that makes dense supervision possible.