Study on Face Direction Measurement and Object Detection for Gazed Object Estimation
Nanami Saito, Hiroshi Takemura, Hiroshi Mizoguchi · The Proceedings of JSME annual Conference on Robotics and Mechatronics (Robomec) · 2019
Service robots that support people need a function to estimate gazed object. Conventional research has estimated gazed object by measuring face direction and using input data beforehand. However, in this method, it can be used only in the state where information is input beforehand, and it can be used only in a limited scene. Therefore, we proposed a method to estimate gazed object using face direction measurement and object detection. We estimated gazed object using two Kinect sensors: for face direction measurement and object detection. We conducted an experiment to ascertain the effectiveness of our proposed method. In the experiment, we estimated gazed object that set at 16 points using the proposal method. As a result, the effectiveness of the proposal method was suggested.