TRAFFIC OFFENDERS PROFILING AND PREDICTION SYSTEM USING A DEEP LEARNING ALGORITHM

Abstract: In developing countries, traffic agencies while trying to discharge their duties end up causing more harm by chasing offenders thereby leading to their doom. This is so because of the absence of an effective and efficient profiling and prediction system that can reduce anonymity of traffic offenders. This thesis … See more

Abstract: In developing countries, traffic agencies while trying to discharge their duties end up causing more harm by chasing offenders thereby leading to their doom. This is so because of the absence of an effective and efficient profiling and prediction system that can reduce anonymity of traffic offenders. This thesis reviews traffic offence systems with a critical observation on the technologies applied and their applicability in both rural and urban areas. From the review, it was discovered that existing traffic offence systems suit urban areas; there was absence of a standard traffic offenders’ profiling system; there was lack of offender’s verification and authentication system in Nigeria thereby shielding most offenders and make them anonymous; there was no efficient communication of traffic situations of an areas, and status of road user and traffic offences between traffic agents and road users/offenders; and the absence of a predictive system to place an offender on a watch list for a probable traffic offence in the future. This study developed a traffic offenders profiling and prediction System with different sub systems that handle offenders profiling, verification and authentication of a road user/offender, communicate traffic situations and other related traffic information amongst traffic agents and road users/offenders, and prediction of probable offence by a road user or traffic offender in the nearest future. The developed system was achieved using a deep learning algorithm of Convolutional Neural Network (CNN) in the design of the predictive module, SMS based technology and Mobile Communication System technology was used in designing the communication sub system, the system interfaced with National Identity Number (NIN) database using fingerprint for verification and authentication of a registrant by linking the traffic database with NIN database. The designed system is web-based with mobile interface. The system was designed using Object Oriented Analysis and Design Methodology (OOADM). The designs were implemented using a web system developed with PHP, MySQL and JavaScript. The software performance was tested using the accuracy of traffic offender prediction. This system will help traffic agents to generate instant traffic history of an offender and create current traffic awareness to road users (vehicle owners and drivers). The result obtained was a web-based system that can profile offenders, communicate traffic related issues via SMS (Short Message Service) and predict the likelihood of a road user committing a traffic blunder in the near future. The system developed shows 95% accuracy of the deep learning technique for prediction. The new system if it is deployed will profile traffic offenders in both urban and rural settings, create a Traffic offender’s database that will interact with existing national databases to authenticate traffic offenders, thereby building the consciousness on every road user that there is no hiding place. Table of ContentsDeclaration iiCertification iiiApproval ivDedication vAcknowledgement viAbstract viiiList of Tables xivTable of Figures xvCHAPTER ONE1.1 Background of the Study 11.2 Statement of the Problem 41.3 Aim and Objectives of the Study 61.4 Significance of the Study 61.5 Scope of the Study 71.6 Limitations of the Study 71.7 Definition of Terms 8CHAPTER TWO2.1 Theoretical Review 92.1.1 Road Traffic 92.1.2 Traffic System 112.1.3 Classification of Traffic System 122.1.4. How to Gather Information for Traffic System 152.1.5 Deep Learning 162.1.6 Data Mining Techniques 202.1.7 Classification Techniques 202.1.7.1 Artificial Neural Networks 212.1.7.2 Decision Tree 222.1.7.3 Bayesian classifiers 232.1.7.4 Clustering Techniques 232.2 Technological Review 242.2.1 Biometric Security 242.2.2 Application of Biometrics in Real-Time System. 282.2.3 Algorithm for Smart Application of Biometrics in Real-Time Transactions 31 2.2.4 Biometrics Processing of Algorithm 322.2.5 Matching Algorithm 322.3 Conceptual Framework 372.3.1 Profiling System 372.3.2 Evaluation of Offender Profiling 402.3.3 Traffic Offenders Profiling System: 422.4 Review of Related Works 442.4.1 Review of Existing Traffic Offenders Profiling System 442.4.2 SMS/MMS-Based Communication System 482.4.3 Authentication System 522.4.4 Review of road traffic offender’s/prediction system 552.4.5 Deep Learning Training Algorithm 582.4.5.1 Deep Learning Algorithm 632.4.5.2 CNN Deep Learning Algorithm 662.5 Summary of Literature Review and Research Gap Established 67CHAPTER THREE3.1 Methodology Adopted 703.2 Method of Data Collection 713.3 Analysis of the Existing System 713.3.1 Data Flow Diagram (DFD) of the Existing System 733.3.2 Advantages of the Existing System 743.3.3 Weakness of the Existing System 743.4 Analysis of the New System 743.4.1 Data Flow Diagram (DFD) of the New System 773.4.2 Advantages of the New System 773.4.3 Use Case Diagram 783.4.4 Sequence Diagram 803.4.5 Flowchart Diagram 813.5 Justification of the New System 823.6 High Level Model of the New System 83CHAPTER FOUR4.1 Objectives of the Design 854.2 Control Centre/Main Menu 854.3 The Submenus/Subsystems 874.3.1 NIN Sub System 874.3.2 Admin Sub System 884.3.3 Traffic Agent Sub System 894.3.4 Police Sub System 904.3.5 Traffic Offenders Sub System 914.4 System Specifications 914.4.1 Database Development Tool 914.4.2 Database Design and Structure 914.4.3 Math Specification 944.4.4 Program Module Specification 954.4.5 Input / Output Format 964.4.6 Algorithm 1024.4.7 Data Dictionary 1044.5 System Flowchart 1054.6 UML Class Diagram 1064.7 System Implementation 1074.7.1 New System Requirements 1074.7.1.1 Hardware Requirements 1084.7.1.2 Software Requirements 1084.7.2 Program Development 1094.7.2.1 Choice of Programming Environment 1094.7.2.2 Language Justification &nbs

See less