EDM25/waymo-ipace-detector-dataset

Waymo I-PACE Vehicle Detection Dataset

YOLO-format object detection dataset for detecting Waymo autonomous vehicles (Jaguar I-PACE) in Austin traffic camera images.

Dataset Structure

├── images/
│   ├── train/     # Training images (JPG)
│   └── val/       # Validation images (JPG)
├── labels/
│   ├── train/     # YOLO format annotations (TXT)
│   └── val/       # YOLO format annotations (TXT)
└── dataset.yaml   # YOLO configuration

Label Format

YOLO normalized coordinates:

class_id center_x center_y width height
0 0.512 0.425 0.284 0.156

Usage with Ultralytics YOLO

from huggingface_hub import snapshot_download
from ultralytics import YOLO

# Download dataset
dataset_path = snapshot_download(
    repo_id="EDM25/waymo-ipace-detector-dataset",
    repo_type="dataset",
    local_dir="./dataset"
)

# Update dataset.yaml path for local use
import yaml
with open(f"{dataset_path}/dataset.yaml", "r") as f:
    config = yaml.safe_load(f)
config["path"] = dataset_path
with open(f"{dataset_path}/dataset.yaml", "w") as f:
    yaml.dump(config, f)

# Train model
model = YOLO("yolo11s.pt")
model.train(data=f"{dataset_path}/dataset.yaml", epochs=100)

Source

Images collected from Austin, TX public traffic cameras via the Austin Mobility API (https://data.mobility.austin.gov/).

License

MIT License