發(fā)布時(shí)間：2018-10-08 16:48

【摘要】：為了解決復(fù)雜場(chǎng)景下手勢(shì)識(shí)別的問(wèn)題,將基于區(qū)域的全卷積網(wǎng)絡(luò)(R-FCN)用于手勢(shì)識(shí)別.為了使網(wǎng)絡(luò)適應(yīng)復(fù)雜場(chǎng)景,利用在線難例挖掘技術(shù)對(duì)手勢(shì)識(shí)別過(guò)程中產(chǎn)生的難例進(jìn)行在線學(xué)習(xí),并結(jié)合手的特征對(duì)網(wǎng)絡(luò)參數(shù)進(jìn)行優(yōu)化調(diào)節(jié).實(shí)驗(yàn)結(jié)果表明:基于R-FCN的手勢(shì)識(shí)別方法能準(zhǔn)確地從復(fù)雜場(chǎng)景中識(shí)別手勢(shì),識(shí)別率達(dá)到99.73%.
[Abstract]:In order to solve the problem of hand gesture recognition in complex scenes, a region based full convolution network (R-FCN) is applied to gesture recognition. In order to adapt the network to the complex scene, the online hard case mining technique is used to study the difficult cases in the process of hand gesture recognition, and the network parameters are optimized and adjusted with the hand features. The experimental results show that the gesture recognition method based on R-FCN can recognize gestures accurately from complex scenes, and the recognition rate is 99.73.
【作者單位】： 華中科技大學(xué)自動(dòng)化學(xué)院;華中科技大學(xué)圖像信息處理與智能控制教育部重點(diǎn)實(shí)驗(yàn)室;
【分類號(hào)】：TP391.41

【相似文獻(xiàn)】

相關(guān)期刊論文 前10條

1 武霞;張崎;許艷旭;;手勢(shì)識(shí)別研究發(fā)展現(xiàn)狀綜述[J];電子科技;2013年06期

2 ;新型手勢(shì)識(shí)別技術(shù)可隔著口袋操作手機(jī)[J];電腦編程技巧與維護(hù);2014年07期

3 任海兵,祝遠(yuǎn)新,徐光，

本文編號(hào)：2257519