A Comprehensive Review on Real-Time Hand Gesture Recognition using Computer Vision and Deep Learning

Abstract

Hand gesture recognition has become an important area of research in human-computer interaction (HCI) because it lets people and computers talk to each other in a natural way without touching anything. Recent progress in computer vision, machine learning, and deep learning has greatly improved the speed and accuracy of real-time gesture recognition. This has made it useful in many fields, including virtual and augmented reality, immersive gaming, healthcare monitoring, assistive technologies for people with disabilities, and smart robotic control.
This review brings together all the research that has been done on real-time hand gesture recognition. It looks at both traditional methods, like template matching and histogram-based descriptors, and newer methods that use Support Vector Machines (SVM), Hidden Markov Models (HMM), Convolutional Neural Networks (CNN), and Recurrent Neural Networks (RNN). OpenCV, TensorFlow, and Google's MediaPipe are among of the current frameworks that are given special attention. These frameworks have made it easier to create strong implementations on mobile and embedded systems.
Gesture recognition systems still have problems, even though they have come a long way. These problems include changes in lighting, backdrop complexity, occlusion, user-specific characteristics, and privacy issues over biometric data. The comparative analysis in this study elucidates the relative advantages and disadvantages of various algorithms while pinpointing enduring research deficiencies.
Lastly, the paper talks about some promising areas of research, such as multimodal fusion approaches (combining vision with electromyography or depth sensing), lightweight deep learning models that can be used in real time on low-power devices, gesture vocabularies that can be used in different cultures, and advanced uses of sign language translation and AR/VR ecosystems. The goal of these insights is to help create the next generation of gesture recognition systems that are accurate, adaptable, and able to be used in the real world.