Futuristic Finger and its Modern Day Applications

Gesture regulated systems controlled by smart wearables have spearheaded the next epoch in human machine interaction. Gestures being intuitive and expressive are a more convenient way of communication. However, for developing a gesture controlled system, we need to accurately detect fingertips. In this paper, we present a fingertip detection system that can be efficiently used by smart wearables. This approach is free of markers and centroid-based techniques which are traditionally used to detect fingertips.The system's functionality is controlled by the number of fingertips in the frame. We collated a customized dataset, `1-2-3-4 Hands', which contained the images of different hands gesturing using one to four fingers. Using Faster RCNN with Inception v2 module, we trained this dataset to build a model capable of recognizing any of the first four fingertips, excluding the thumb. The count of fingertips is used to perform an action in real-time gesture controlled systems. Finally, we have implemented finger-control solutions such as AirWriting, AirDrawing, and Gaming Controls and enlisted their benefits.