International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 12 Issue: 12 | Dec 2025
p-ISSN: 2395-0072
www.irjet.net
The Role of Mobile Technology in Collecting Traffic Data for Object Detection: A Framework Tailored for the Gambia Moses Correa1, Moses W. Nyenkan2, 1, Department of Mechanical and Power Engineering Henan Polytechnic University, Henan, China. 2, Material Science and Engineering Henan Polytechnic University, Henan, China.
-------------------------------------------------------------------------------***------------------------------------------------------------------
Abstract - The sophisticated advancement of mobile
popular. This has turned them into distributed sensing nodes that can get useful information via built-in GPS, cameras, accelerometers, and communication components (Aloqaily et al., 2020). This convergence of technologies has led to new digital solutions for traffic monitoring, road safety assessment, and object recognition, all of which are necessary for sustainable urban planning and reducing traffic jams. The GSMA (2024) says that mobile technology is growing quickly in Sub-Saharan Africa, which has more than 500 million unique users and 62% of people using smartphones. More precisely, mobile crowd-sourcing projects can be started in The Gambia, where more than half of the people can now get online. This is a good place to start (Data Reportable, 2024). This ease of access makes it possible for individuals to collect data in real time on their mobile devices. This means that average people act as human sensors and help the country regulate traffic (Good child, 2007). Crowd-sourced mobile data, aggregated and refined using artificial intelligence (AI) pipelines, enhances object detection models by supplying diverse, contextual, and localized imagery (Liu et al., 2021). Localized data can help machine learning algorithms find the distinctive patterns of traffic, informal transportation, and road conditions when resources are limited (Abdullah et al., 2022). Free-flowing traffic data gathering methods, such loop detectors, surveillance cameras, radar sensors, and others, cost a lot of money to buy and keep up with, which is too much for many developing nations (Banerjee et al., 2020). On the other hand, mobile technologies are a cheaper and more flexible option that lets you record multi-modal data practically all the time, both in cities and in the country. Researchers have shown that cellphones can give reliable information about things like foot and car movement and anything strange on the road (Eren et al., 2012) by using built-in sensors like accelerometers and gyroscopes. These data streams, when combined with AI-based analytics, make it possible to find objects and events almost in real time, which helps keep traffic and infrastructure safe (Zhang et al., 2023). Annotated datasets utilized for the development of object detection algorithms must possess high quality and accurately reflect realistic driving scenarios (Redmon and Farhadi, 2018). However, most of the datasets that are accessible, such KITTI or COCO, are generally biased toward data from high-income countries. This can lead to performance biases when the data is used in Africa (Kaggle et al., 2022). Pedestrians, bikers, minibuses, donkey carts, and animals are examples of typical road users in The Gambia, which are poorly represented in worldwide datasets. Taking advantage of the mobile technology to record scenes of local
technology has provided novel opportunities for real-time traffic information gathering and intelligent transportation systems, particularly in resource-constrained environments such as The Gambia. The study analyzes the utilization of cellphones as compact decentralized data collecting devices and their role as the basis for context-specific object detectors employing the aforementioned sensors (cameras, GPS, and accelerometers). The research generated the Gambian Traffic Object Dataset (GTOD v1), comprising over 178,000 annotated instances across seven object classes, featuring region-specific attributes like livestock and donkey carts, acquired via a mixedmethod approach involving participatory mobile sensing, deep learning, and federated learning. Upon fine-tuning the dataset with the YOLOv5-based detection model, the average mean Average Precision (mAP) at 0.5 was 0.83, enabling real-time inference at an average of 27 frames per second on mobile hardware. They reduced bandwidth consumption by 72 percent and significantly enhanced data privacy with minimal accuracy loss via federated learning. The spatial analysis of the heat map identified congestion hot spots and mobility patterns, which can be utilized in urban planning and road safety policy formulation. These results confirm that in developing nations, mobile technology provides a cost-effective, privacy-conscious, and scalable platform for the development of intelligent, locally adaptive traffic monitoring systems. The research emphasizes that community-based data ecosystems, along with lightweight AI architectures, can facilitate both technological advances and ethical governance in transport analytics in Sub-Saharan Africa. Key Words: Mobile sensing, object detection, federated learning, traffic data, YOLOv5, deep learning, The Gambia, participatory data collecting, intelligent transportation systems, GIS-based analytics.
1. INTRODUCTION Mobile technology has improved intelligent transportation systems (ITS), which collect, analyze, and use traffic data to help cities manage traffic. Mobile phones offer an unparalleled opportunity to gather substantial quantities of real-time and traffic data from diverse objects within a confined area, particularly in developing nations such as The Gambia, where the establishment of permanent sensor networks and other surveillance infrastructure is limited. Smartphones are becoming more and more important, especially because the mobile internet is becoming more
© 2025, IRJET
|
Impact Factor value: 8.315
|
ISO 9001:2008 Certified Journal
|
Page 623