Top-View Vehicle Detection Dataset
Annotated dataset of 626 images seen from above, intended for vehicle detection via YOLOv8. Ideal for traffic analysis, urban planning, and embedded applications.
626 images (640x640), 1,257 files in total with YOLOv8 annotations
CC BY 4.0
Description
This dataset contains 626 images annotated in a top-view view, capturing various types of vehicles (cars, trucks, buses). Each image is preprocessed at a resolution of 640x640 and accompanied by annotations in YoloV8 format. The data set is structured into training sets (536 images) and validation sets (90 images), with increases applied to the training set.
What is this dataset for?
- Train object detection models for vehicle recognition
- Test architectures like YoloV8 on aerial views in a real environment
- Prototype smart urban mobility or traffic surveillance solutions
Can it be enriched or improved?
Yes, we can enrich this dataset by adding other urban views, additional classes (pedestrians, motorcycles) or by a controlled synthetic increase. It can also be completed by embedded sensors (LIDAR, GPS) for multimodal cases.
🔎 In summary
🧠 Recommended for
- Embedded vision engineers
- AI student projects
- Smart mobility prototypes
🔧 Compatible tools
- YoloV8
- Roboflow
- Ultralytics
- OpenCV
- FiftyOne
💡 Tip
To improve robustness, combine this dataset with other angles or sensors (dashcam, drone, etc.)
Frequently Asked Questions
Does this dataset work with Yolov8 directly?
Yes, the annotations are already formatted for Yolov8, ready to be used without conversion.
Does this dataset contain multiple object classes?
No, it focuses only on the “vehicle” class, but this specialization reinforces the quality of targeted detection.
Is the volume sufficient for a production model?
It is ideal for prototyping or fine-tuning, but may require additional data for large-scale deployment.
