Hand Detection and Gesture Recognition in Complex Backgrounds

In this paper, a Convolutional Neural Networks (CNN) based hand detection model that, on a major note, focuses and segments only the hands from any complex background using the Open-CV libraries for real-time computer vision, is proposed. Based on the features extracted from the region of interest, the VGG16 CNN Architecture classifies and predicts the gestures, based on the trained data. The system is trained by using binary images, so that the background is eliminated and classification is done only on the edges. This approach increases the performance of the system with respect to time. The major step involved in the proposed system is Background Elimination, which is carried out using a series of Open-CV methods and functions. Hand Detection Systems find applications in various domains ranging from Sign-Language Detection to Human-Computer Interaction.