Classification of Human Fall from Activities of Daily Life using Joint Measurements

摘要: Falls are a major health concern to most of communities with aging population. There different approaches used in developing fall detection system such as some sort wearable, non-wearable ambient sensor and vision based systems. This paper proposes using Kinect for Windows generate depth stream which is classify human from other activities daily life. From the experimental results our was able achieve an average accuracy 94.43% sensitivity 94.44% specificity 68.18%. The also showed that brutally sitting on floor has higher acceleration, very close acceleration shown by fall. Even then high determining brutal movements use joint positions, this indication further improvements algorithm can make more robust.