Interactive Context-Aware Furniture Recommendation using Mixed Reality

We present a Mixed Reality (MR) system, through Hololens, to provide context-aware furniture recommendation in an interactive fashion. Firstly, a ranking-based metric learning method is adopted to represent the furniture compatibility through a latent space. Then, in the recommendation process, a physical scene is captured by the cameras mounted on the MR device, and two types of scene context are analyzed: (1) category context; (2) spatial context. At last, the one with the minimal weighted ranking distance in the latent space is recommended to the user. With MR devices, a user could perceive and manipulate the recommended furniture in real-time. We conduct user study to validate the efficacy of the proposed system.