Abstract:
In the era of increasing urbanization and traffic congestion, efficient traffic man-
agement systems are crucial for ensuring smooth vehicular flow and enhancing road
safety. This project presents a extensive study and implementation of a Bangladeshi
Vehicle Classification and Detector using Deep Learning and Convolutional Neural
Networks (CNN).
The primary objective of this project is to develop a robust system capable of
classifying vehicles commonly found on the roads of Bangladesh and detecting their
presence with high accuracy. The system harnesses the power of deep learning,
specifically CNNs, to achieve this goal.
The project involves a multi-stage process, beginning with data collection and
pre-processing. A diverse data-set of Bangladeshi vehicles is curated and annotated.
Subsequently, a deep neural network is designed and trained using popular modules
to classify these vehicles into various categories, such as cars, motorcycles, buses,
and rickshaws. The renowned CNN model used is optimized through rigorous ex-
perimentation of public effort to ensure optimal performance.
Furthermore, a real-time vehicle detection module is integrated into the system,
enabling it to identify and locate vehicles within images or video streams. This
detection capability can be invaluable for traffic monitoring, surveillance, and smart
city applications. The results obtained from the project demonstrate the effective-
ness of deep learning and CNNs in vehicle classification and detection tasks specific
to the Bangladeshi context. The system exhibits promising accuracy rates and
real-time capabilities, making it a valuable tool for traffic management and safety
enhancement.
This project contributes to the advancement of intelligent transportation systems
in Bangladesh and provides a foundation for future research in the field of computer
vision and deep learning for traffic-related applications.
