Different Categories of Feature Extraction and Machine Learning Classification Models Used for Hand Gesture Recognition Systems: A Review

Abstract

Hand gestures are among the most common aspects of body expression that can be used for interactive communications. Hand movements can convey language information. Since they can facilitate interaction and provide a practical paradigm of interaction that can be applied in a number of situations, hand gestures are a fascinating issue for research. The HGR (Hand Gesture Recognition) system is used to recognise hand gestures. Feature extraction and classification are two stages that are crucial to the fundamental architecture of hand gesture recognition. In a hand gesture recognition system, feature extraction refers to techniques for establishing connections between two variables and accurately representing the information. The classifiers are used to classify the features of the images. Several feature extraction techniques and classifiers are mentioned in this paper. Hand gesture recognition is extensively used in SL (sign language), TV controlling, HCI (human computer interaction) systems, VR (virtual reality), games, etc. due to its creativity and rapidity of involvement. The various existing techniques for HGR are surveyed and analyzed. The analysis of existing techniques showed that there are some major issues with existing techniques, such as data augmentation and generalisation issues, reliability issues, limited datasets, Complex backgrounds degrade the performance of the existing techniques. The diversity and complexity of actions have a massive effect mostly on classification performance and accuracy throughout the purpose of selecting hand movements. In order to increase the accuracy of hand gesture recognition, machine learning methods can be applied. In this paper, various existing techniques are surveyed and compared with different parameters such as accuracy, precision, sensitivity, specificity, etc.