Indonesian Journal of Electrical Engineering and Computer Science Vol. 13, No. 2, February 2019, pp. 787~793 ISSN: 2502-4752, DOI: 10.11591/ijeecs.v13.i2.pp787-793  787 Journal homepage: http://iaescore.com/journals/index.php/ijeecs AI-Based Targeted Advertising System Tew Jia Yu, Chin Poo Lee, Kian Ming Lim, Siti Fatimah Abdul Razak Faculty of Information Science and Technology, Multimedia University, Malaysia Article Info ABSTRACT Article history: Received Oct 15, 2018 Revised Dec 16, 2018 Accepted Dec 30, 2018 The most common technology used in targeted advertising is facial recognition and vehicle recognition. Even though there are existing systems serving for the targeting purposes, most propose limited functionalities and the system performance is normally unknown. This paper presents an intelligent targeted advertising system with multiple functionalities, namely facial recognition for gender and age, vehicle recognition, and multiple object detection. The main purpose is to improve the effectiveness of outdoor advertising through biometrics approaches and machine learning technology. Machine learning algorithms are implemented for higher recognition accuracy and hence achieved better targeted advertising effect. Keywords: Age estimation Gender recognition Object recognition Targeted advertising Vehicle recognition Copyright © 2019 Institute of Advanced Engineering and Science. All rights reserved. Corresponding Author: Chin Poo Lee, Faculty of Information Science and Technology, Multimedia University, Jalan Ayer Keroh Lama, 75450 Bukit Beruang, Melaka, Malaysia. Email: cplee@mmu.edu.my 1. INTRODUCTION Billboard advertising is a type of Out-Of-Home advertising that grabs the chances of outdoor promotion which typically achieves desirable results. In contrast to traditional billboards with static messages, digital billboards with more flexibility and up-to-date messages are taking their way of replacing them. This more advanced form of advertising is known as the Digital Out of Home (DOOH). As technology evolved, digital advertising becomes increasingly popular not only because of its tendency of lower cost, but also its targeting and interactive features with the use of cameras, sensors and other add-on devices. Digital billboards are called “smart” or “intelligent” with their capabilities of recognizing a particular object and display relevant content to it. These billboards are connected to devices for collecting inputs, and a system is working behind as the control. Along with all the algorithms and processing functions, the billboard will be targeting a certain group and display relevant advertisement for better attention and influences. Several issues are identified in the existing advertising systems as the following: (1) unsuitable advertisements are displayed to the audience, (2) unable to target outdoor audience without any activity, and (3) limited functionalities. These problems result in wastage of resources, high costs in advertising, and ineffective advertising. In view of this, an intelligent targeted advertising system is proposed for the following objectives: (1) display better-targeted advertisements to the audience, (2) improve the effectiveness of outdoor advertising, and (3) offer wide range of functionalities in a single system. The proposed system is capable of recognising gender and age of detected faces as well as various object categories such as vehicles, electronic devices, and food. Best-suited advertisement relevant to the real-time demographics will be retrieved and displayed on the billboard, solving the problems encountered in existing advertising systems as mentioned before. The rest of the paper is organized as follows. In Section I, an introduction of billboard advertising and targeted advertising is provided. Subsequently, some existing works in smart billboard advertising are