BDRoadSigns: An Image-Based Dataset of Bangladeshi Traffic Signs for Supervised Classification Models

The BDRoadSigns Dataset is a high-quality image collection of traffic signs captured from various regions in Bangladesh, including areas such as 300 Feet Road (Dhaka), Rajshahi, and several other urban and rural locations, between December 23, 2023, and July 10, 2024. There are a total of 18 classes. Now the distribution of per-class images is given below:

1. College Front-302

2. Cross Road-276

3. Drop Off or Pick Up-252

4. Height Limit 5.8-302

5. Left Side Road-300

6. Market Front-282

7. Pass Either Side-300

8. Pedestrian Crossing-283

9. Pedestrian Crossing Ahead-300

10. Rail Crossing-283

11. Rickshaw Restrictioned-282

12. Right Turn-300

13. RoundAbout-283

14. Side road Left-300

15. Speed Limit 40km-300

16. Turn Left-300

17. Turn Right-190

18. U-Turn-300

The images were captured using a Realme 6i smartphone under natural lighting conditions, ensuring high resolution and clarity. Originally taken in JPG format at 4000 × 3000 pixels, the images were resized to 480 × 640 pixels at 72 DPI to ensure consistency across the dataset. While the dataset does not include annotations such as bounding boxes or masks, it serves as a valuable resource for researchers and deep learning practitioners working on traffic sign classification. This dataset is especially relevant for applications in intelligent transportation systems, autonomous driving, and road safety analysis within the context of Bangladeshi road environments.