A hardware accelerator for dynamic gesture recognition in the field of human-computer interaction

Abstract

The invention belongs to the technical field of integrated circuits, in particular to a dynamic gesture recognition hardware accelerator oriented to the field of human-computer interaction. The circuit of the present invention includes: an image preprocessing module, a standby module, a RISCV processor, and a dynamic gesture recognition coprocessor; the accelerator adopts a low-complexity gesture recognition algorithm to convert RGB images into YCrCb images, and extract binary skin color according to the YCrCb images image, effectively reducing the on-chip storage capacity, reducing the power consumption of data transfer, and thus improving the energy efficiency ratio of the entire chip; the accelerator uses the peripheral contour to find the center of gravity of the gesture, and obtains the direction of gesture movement by tracking the displacement of the center of gravity, which has excellent recognition effect and can be widely used It is used in the field of non-contact human-computer interaction; in addition, the accelerator is controlled by a RISCV processor, which has strong flexibility, and a standby module is set for long-term standby operation, effectively reducing the power consumption of long-term work.

Description

technical field [0001] The invention belongs to the technical field of integrated circuits, and in particular relates to a dynamic gesture recognition hardware accelerator oriented to the field of human-computer interaction. Background technique [0002] Natural, barrier-free, highly effective, and contactless new intelligent human-computer interaction systems have become an inevitable trend in information development. As one of the extremely important channels in the field of human-computer interaction, gesture has the advantages of wide application, simple operation, and high frequency of use. Gesture recognition is divided into static gesture recognition and dynamic gesture recognition. Static gesture recognition is classified through static gesture images, such as "ok", "bixin", "yes" and other gestures. Dynamic gesture recognition is through gestures under video sequences Behavior analysis, such as "up", "waving", "push" and other gestures. However, for dynamic gestur...

AI Technical Summary

Problems solved by technology

However, for dynamic gesture recognition, traditional algorithms based on template training, feature extraction, gesture classification, etc. have too high algorithm complexity, high resource requirements, and correspondingly high power consumption, which cannot be deployed on mobile terminals, IoT , in wearable devices

In addition, due to the high complexity of the external environment, the practical application of intelligent algorithms such as machine learning in the field of dynamic gesture recognition is facing great challenges, and the number of parameters implemented by intelligent algorithms is large and complex, and it is also difficult to meet the needs of mobile terminals, IoT, etc. , Wearable devices require low power consumption

Images