一种低时间消耗的手机位置自适应动作识别方法
首发时间:2014-11-15
摘要:基于智能手机的人体动作识别在用户行为跟踪、个性化推荐等领域有着广泛的应用。实现手机在不同位置的自适应是该领域需要解决的关键问题。本文提出了一种低时间消耗的手机位置自适应的动作识别方法。首先文章提出使用连续动作作为重训练数据。然后通过建立人体运动的隐马尔可夫模型分析了连续动作识别正确的概率,为重训练提供理论依据。最后通过实验验证了方法的正确性。实验表明,该方法可以提高1.2%自适应识别率情况下,降低84.3%时间消耗。
For information in English, please click here
A low time consuming phone location adaptive activity recognition method
Abstract:Smartphone based human activity recognition is widely used in user behavior tracking, personalized recommendations and other fields. Implement adaptive mobile phones in different locations is a key issue to be resolved in this area. It proposes a low time consumption method for phone location adaptation activity recognition. First, the continuous recognition result is used for retraining. Then, it analyses the probability of continuous results through establishing hidden Markov model. Finally, it verifies the method through experiments. The experiments show that our method can improve the recognition rate about 1.2%, and reduce the time consumed about 84.3%.
Keywords: mobile computing activity recognition hidden Markov model
论文图表:
引用
No.4616884101402714****
同行评议
共计0人参与
勘误表
一种低时间消耗的手机位置自适应动作识别方法
评论
全部评论0/1000