A Study on 3D Gaze Recognition from the Environment Side by Multiple Optical Sensors

There is a high demand for information on what people are looking for when shopping and making decisions. Although wearable eye tracking devices can provide good gaze estimation results, it is not practical to ask customers to wear a wearable type in the store. Therefore, we proposed a general, accurate pipeline for tracking the gaze of people in the 3D space by integrating several different optical sensors on the environment side. The accuracy of the system is comparable to wearable devices so that we successfully distinguished test points that are only 30cm apart on the shelf. The pipeline also can be used to restore the purchase process and applied to a variety of tasks by updating those separated parts.

There is a high demand for information on what people are looking for when shopping and making decisions. Although wearable eye tracking devices can provide good gaze estimation results, it is not practical to ask customers to wear a wearable type in the store. Therefore, we proposed a general, accurate pipeline for tracking the gaze of people in the 3D space by integrating several different optical sensors on the environment side. The accuracy of the system is comparable to wearable devices so that we successfully distinguished test points that are only 30cm apart on the shelf. The pipeline also can be used to restore the purchase process and applied to a variety of tasks by updating those separated parts.