Object Detection in Bangladeshi Road Conditions

Overview:

This project focuses on developing solutions for object detection in various road conditions across Bangladesh. The dataset consists of 9825 images captured under real-world conditions using smartphone cameras, covering diverse road types such as towns, expressways, highways, and village roads. Annotations are provided for 13 classes including auto rickshaws, bicycles, cars, and pedestrians.

Dataset Description:

Total Images: 9825

Total Annotated Objects: 78,943

Classes: auto_rickshaw, bicycle, bus, car, cart_vehicle, construction_vehicle, motorbike, person, priority_vehicle, three_wheeler, train, truck, wheelchair

Data Collection:

All images were collected to simulate real-world driving scenarios in Bangladesh, ensuring authenticity and practical applicability for autonomous vehicles.

File Structure:

train.csv: CSV file containing image IDs and corresponding prediction strings.

images: Directory containing the dataset images.

Usage:

Clone the repository.
Install the necessary dependencies.
Load and explore the dataset using Python or your preferred data analysis tool.
Develop and train object detection models using the provided dataset.

Acknowledgements:

This dataset was collected and annotated as part of the efforts to advance object detection technology for safer driving experiences in Bangladesh.