| Summary: | Recently, gesture recognition technology has attracted increasing attention because it provides another means of information exchange in some special occasions, especially for auditory impaired individuals. At present, the fusion of sensor signals and artificial intelligence algorithms is the mainstream trend of gesture recognition technology. Therefore, this article designs a machine-learning-empowered gesture recognition glove. We fabricate a flexible strain sensor with a sandwich structure, which has high sensitivity and good cycle stability. After the sensors are configured in the knitted gloves, the smart gloves can respond to different gestures. Additionally, according to the representation characteristics and recognition targets of sampled signal data, we explore a segmented processing method of dynamic gesture recognition based on Logit Adaboost algorithm. After classification training, the recognition accuracy of smart gloves can reach 97%.
|
|---|