Fingertip detection for air-writing applications

The purpose of this work is the development of an air-writing system based on fingertip detection using deep learning and image processing techniques. This system consists of a convolutional neural network that is capable of automatically learning the characteristics related to the appearance of the fingertip, and also locating it by solving a regression problem.

A color camera has been to acquire the user's movements. The proposed system can locate the fingertip with great precision by estimating its X and Y coordinates in the image. This system can be used to develop Human-Computer interfaces more natural and friendly, allowing to perform various functions more comfortably, such as surfing Internet more easily, or even facilitating the tasks of home automation (changing the temperature, etc.)