Camera-Based Smart Lighting System that complies with Indonesia’s Personal Data Protection Act

Cameras and image processing can detect when a user is present and perform smart lighting control according to the image processing results to obtain electricity savings. Several studies have implemented smart lighting cameras that protect privacy using low-res cameras. However, to comply with the general data protection regulation (GDPR), the system must comply with other requirements. This paper proposes a camera-based smart lighting system that complies with Indonesia’s Personal Data Protection Act. First, we explain the concept of a legacy camera-based smart lighting system. Then we explain the requirements demanded by the GDPR. After that, we developed a new camera-based smart lighting architecture that complies with these requirements, including implementing data perturbation, image pixelation, and a convolutional neural network (CNN) model. Finally, we benchmarked our architecture with state-of-the-art architectures regarding GDPR compliance. The test results show that with a pixelized image, the accuracy of the test data using our CNN model is 0.75. Some of our perturbed images are correctly detected as persons. Furthermore, this system is refined with a user interface (UI) for consent and a security system provided by Raspberry Pi, which runs the image perturbation. Our contribution is a novel camera-based smart lighting architecture that complies with GDPR requirements.

I. Introduction

Smart lighting is a research area with an increasing trend, especially in its focus on increasing energy efficiency [1]. There are two main strategies for increasing energy efficiency in lighting systems: (1) By dimming the lights when the light is bright enough (2) By turning off the lights when no one is present [2]. Several studies already exist regarding the second strategy, one of which is with the camera and image processing which can detect when a user is present and exercise control according to the image processing results [3]. However, there is a concern when using the camera in smart lighting, namely whether the use of the camera violates privacy or not.