Safe Interaction with Machines via Hand Gesture Recognition in Public Areas

Till now many customized human-machine contactless interfaces have been designed, but the outburst of coronavirus pandemic has made us realized to develop general purpose contactless machine interface that can be used in public area. The most promising solution is vision-based Hand Gesture Recognition (v-HGR) as compared to voice-based interfaces. Simultaneously, due to high flexibility in hand geometry and constrain in the level of technical understanding of general people, it is becoming popular. Many use-cases in real-life scenarios are included in this paper that can improve health environment and can bring a great relief to the entire society. In this paper, a model combining Faster R-CNN with SqueezeNet and R-CNN is used to detect and recognize hand gestures which has achieved an accuracy of 99.0% which is better than existing techniques working with RGB images.