Lightweight gesture recognition method based on deep learning

Abstract

The invention discloses a lightweight gesture recognition method based on deep learning, and belongs to the technical field of artificial intelligence. The method comprises the following steps: in a hand detection stage, providing an improved MobileNetv2-YOLOv3 network structure, and greatly reducing the parameter quantity and the calculation quantity of a model while ensuring the accuracy; in the hand key point detection stage, an attention-based selective gesture distillation method (Attention-based Selective HAND Distillation, ASHD) is provided, firstly, a knowledge network (Teaser Model, T) and a lightweight basic network (Student Model, S) which are high in expression ability and large in parameter quantity are designed, then structured knowledge of the knowledge network is selectively migrated through the attention-based gesture distillation method, and the hand key point is detected through the Attention-based Selective HAND Distillation method. Jointly training a basic network with a small number of parameters in combination with a real label; in the gesture classification stage, ResNet is adopted as a basic network, and a series of tricks are combined to improve the generalization ability of the model. The method is used for designing a lightweight model, the calculation amount is reduced while the accuracy is guaranteed, and the method can be deployed on embedded equipment with low calculation power requirements.

Description

technical field [0001] The patent of the present invention belongs to the technical field of artificial intelligence, and particularly relates to a lightweight gesture recognition method based on deep learning. Background technique [0002] As a key research direction in the computer field, human-computer interaction technology has made great progress with the development of deep learning and other technologies. Gestures have rich expressive abilities and have great application prospects and value. Gesture recognition technology has strong application prospects in virtual reality, robotics, smart home and other fields. The patent "Gesture Recognition Method, System and Gesture Recognition Gloves Based on Gesture Recognition Gloves" completely extracts gestures, but gesture recognition gloves limit the expression of more gestures and rely on data sensors in the gloves in contact with human hands The patent "A Deep Learning-Based Gesture Recognition Method and Device" propos...

AI Technical Summary

Problems solved by technology

The patent "Gesture Recognition Method, System and Gesture Recognition Gloves Based on Gesture Recognition Gloves" completely extracts gestures, but gesture recognition gloves limit the expression of more gestures and gestures and rely on sensors in the data glove in contact with human hands Sensitivity; the patent "A Method and Device for Gesture Recognition Based on Deep Learning" proposes a complete process of gesture recognition, but without compressing the model, it cannot meet the real-time detection requirements

Images