Thesis Examination Committee

Prof Danny TSANG, ECE/HKUST (Chairperson)
Prof James SHE, ECE/HKUST (Thesis Supervisor)
Prof Wai Ho MOW, ECE/HKUST

Abstract

Interactive dance is a new form of dance in which there is a two-way communication between dancers and visual backgrounds through sensing devices. Visual backgrounds are very important for interactive dance performances to enhance the audience experience. However, producing visual backgrounds is labor intensive and technically demanding. Therefore, the first challenge is how to help dancers get the visual background images more easily. On the other hand, current sensing devices exist limitations on tracking dancers’ distance, the most evident stage information. Therefore, the second challenge is how to overcome the limitations of current sensing devices and enable the interaction based on the distance.
     
This thesis proposes a visual background analytics and triggering system to enable dancers to design and interact with the visual backgrounds. To solve the first challenge, the system incorporates visual background analytics with a recommendation engine. The core of the recommendation engine is a deep matrix factorization (DMF) model which recommends dance background images to dancers by considering the object feature, style feature, and dancers' rating information simultaneously. To solve the second challenge, the system incorporates a Bluetooth Low Energy (BLE) beacon-based triggering engine. The engine leverages a BLE beacon body area network and a smartphone to monitor a dancer's relative distance. The effectiveness of the proposed methods has been proven through a series of experiments. Additionally, a live performance is conducted to demonstrate the practicability of the proposed system.