Augmented Reality (AR) can seamlessly create an illusion of virtual elements blended into the real world scene, which is one of the most fascinating human-machine interaction technologies. AR has been utilized in a variety of real-life applications including immersive collaborative gaming, fashion appreciation, interior design, and assistive devices for individuals with vision impairments. This paper contributes a real-time AR application, ARFurniture, which will allow the users to envision furniture-ofinterests in different colors and different styles, all from their smart devices. The core software architecture consists of deep-learning based semantic segmentation and fast-speed color transformation. Our software architecture allows the user prompt the system to colorize the style of the furniture-of-interest within the scene on their mobile devices, and has been successfully deployed on mobile devices. In addition, using eye gaze as a pointing indicator, a head-mounted user-centric augmented reality based indoor decoration style colorization concept is discussed. Related algorithms, system design, and simulation results for ARFurniture are presented. Furthermore, a no-reference image quality measure, Naturalness Image Quality Evaluator (NIQE), was utilized to evaluate the immersiveness and naturalness of ARFuniture. The results demonstrate that ARFurniture has game-changing value to enhance user experience in indoor decoration.