At present, limited hand gesture databases are made available for reference and research, and these databases are small in scale with limited variants of gestures. In response to this problem, we have established a large-scale dynamic gesture database called LR-DHG database. The sensors for collecting data include Leap Motion and RealSense. The data formats include video frame, depth images, color images, and hand-joint point coordinates. In this paper, we describe in detail the recording work of this database, and use subjective evaluation methods to conduct preliminary tests on it, resulting in a higher recognition rate. This database provides more valuable reference and research data for human-computer interaction.
展开▼
