RSD-BD: Road Surface Damage Image Dataset from Major Cities of Bangladesh for Deep Learning & Computer Vision Research

This dataset comprises 1500 high-quality images depicting various forms of road surface damage collected from major cities in Bangladesh, specifically Dhaka, Mymensingh, and Chattogram. The dataset captures real-world conditions of urban and semi-urban road networks, providing valuable visual data for analysis in computer vision, machine learning, deep learning, and civil infrastructure research. The images are captured under diverse lighting conditions and angles, ensuring variability and practical utility for robust algorithm development.

Dataset Composition:

Total Images: 1500

Format: JPG

Image Resolution: Varied, high-resolution suitable for computer vision tasks.

Class-wise Distribution:

Asphalt Damage: 500 images

Crack: 500 images

Pothole: 500 images

Dataset Potential Applications:

Training, validation, and benchmarking for deep learning and machine learning algorithms focusing on road infrastructure assessment.

Development of computer vision-based automated systems for road damage detection and classification.

Research and development in intelligent transportation systems (ITS), smart city infrastructure management, and predictive road maintenance.

Analysis and testing of algorithms for damage severity assessment and automated cost estimation for repairs.

Intended Users:

Researchers in civil engineering, transportation, and urban planning.

Machine learning and computer vision practitioners focus on infrastructure monitoring and predictive maintenance.

Government bodies and policymakers are interested in infrastructural health assessments and proactive maintenance planning.