Non-cooperative hand detection method

The invention discloses a non-cooperative hand detection method, which comprises four processes of extracting features of an input image through a deep neural network, then performing feature enhancement through a context information attention module, then inputting a region suggestion network to ex...

Full description

Saved in:
Bibliographic Details
Main Authors GUO ZHENHUA, XIE ZHIHUAI
Format Patent
LanguageChinese
English
Published 07.06.2019
Subjects
CALCULATING
COMPUTING
COUNTING
HANDLING RECORD CARRIERS
PHYSICS
PRESENTATION OF DATA
RECOGNITION OF DATA
RECORD CARRIERS
Online AccessGet full text

Cover

More Information
Summary:The invention discloses a non-cooperative hand detection method, which comprises four processes of extracting features of an input image through a deep neural network, then performing feature enhancement through a context information attention module, then inputting a region suggestion network to extract suggestion boxes, and finally performing classification and regression of the suggestion boxes. Finally, an end-to-end deep neural network for hand detection is trained, the network has a hand detection function, and a detection result can be obtained as long as an image is input into the network. 本发明公开了一种非配合手部检测方法，包括四个过程：输入图像经过深度神经网络提取特征，接着经过上下文信息注意模块进行特征增强，再输入区域建议网络提取建议框，最后进行建议框的分类与回归。最终训练出一个手部检测的端到端深度神经网络，该网络就具备检测手的功能，只要对网络输入图像就可以得到检测结果。
Bibliography:Application Number: CN201910114857