Abstract:
In this thesis, the research is mainly focused on traffic object detection by different computer vision algorithm like MobileNetV2 and YOLOv5. The main focus is on real-time detection and identification of various types of automobiles specific to Bangladesh. Vehicle detection is a necessary step for traffic surveillance and autonomous vehicles. Vehicle counting in complex transportation conditions requires the detection and tracking of mobile vehicles. Vehicle detection on the road are used for vehicle tracking, vehicle traffic assessment, average velocity of each individual vehicle, motion analysis, and vehicle classification, and may be applied in a variety of contexts. Because of the irregular traffic, the variety of vehicles, and the absence of a good dataset, it is more difficult in Bangladesh to adopt a smart traffic. One of the most important aspects of algorithm training is data quality. Here, the dataset, “Dhaka Traffic Detection Challenge Dataset” was cleaned and augmented to get better results. The dataset was trained on two neural network architecture, YOLOv5 and MobileNetV2. YOLOv5 ran on PyTorch and the model was trained on Google Colaboratory, a cloud-based platform. Codes were written in Python 3.8 and Python 3.9. Roboflow, an online-based computer vision application, was used to organize the dataset for training. For image categorization and mobile vision, MobileNet is built on the CNN architectural model. There are other models available, but MobileNet is unique in that it uses very minimal compute resources to operate and also applies transfer learning. As a result, MobileNet is ideal for both mobile devices and web browsers. There are 28 levels in MobileNet. MobileNet contains 4.2 million parameters by default. YOLOv5 performed better than MobileNetV2, in terms of accuracy and inference time. This research will be a vital step towards intelligent traffic detection system that can detect unauthorized vehicles like rickshaw/CNGs in highways, or to develop a traffic plan that minimizes traffic congestion on the road.
Description:
Supervised by
Prof. Dr. Golam Sarowar,
Department of Electrical and Electronic Engineering (EEE),
Islamic University of Technology (IUT),
Board Bazar, Gazipur-1704, Bangladesh.
This thesis is submitted in partial fulfillment of the requirements for the degree of Bachelor of Science in Electrical and Electronic Engineering, 2022.