Road Vehicle Images Dataset
A dataset of road images taken in Bangladesh, designed for vehicle detection via YoLov5, including ready-to-use annotations.
6,011 images with annotations, in two train/validation folders, image + YAML formats compatible with YoLoV5
Open Database, Contents: Database Contents
Description
The dataset "Road Vehicle Images" contains over 6,000 images of vehicles on public roads, captured in Bangladesh. Each image is accompanied by an annotation compatible with real-time detection frameworks such as YoLov5. The files are divided between a training folder and a validation folder, facilitating direct use for computer vision projects applied to mobility or road safety.
What is this dataset for?
- Develop vehicle detection models with YoLov5
- Improving traffic recognition for surveillance or autonomous vehicles
- Testing the robustness of models in a real context on urban roads in Asia
Can it be enriched or improved?
Yes. It is possible to enrich this dataset by adding annotations of specific classes (truck, bus, motorcycle...), or by collecting images at night and in bad weather. An extension to other countries or geographical areas would also reinforce the generalization of models.
🔎 In summary
🧠 Recommended for
- YOLO developers
- Smart mobility projects
- Video detection prototyping
🔧 Compatible tools
- YoloV5
- Roboflow
- OpenCV
- Ultralytics Hub
💡 Tip
Combine this dataset with similar games from other countries to train a more generalist detector.
Frequently Asked Questions
Is this dataset already in YoLov5 format?
Yes, images and annotations are directly compatible with YoLov5, including the `data_1.yaml` file for configuration.
Can it be used in a commercial setting?
Yes, the Open Database license allows it, provided you mention the origin and respect the redistribution conditions.
Does this dataset cover several types of vehicles?
Vehicle types are present in the images but are not separated into subclasses. Manual enrichment is possible.
