{"id":624,"date":"2020-06-14T15:39:41","date_gmt":"2020-06-14T06:39:41","guid":{"rendered":"https:\/\/kmlab.nagaokaut.ac.jp\/?page_id=624"},"modified":"2020-06-14T15:39:41","modified_gmt":"2020-06-14T06:39:41","slug":"walkposture_detection","status":"publish","type":"page","link":"https:\/\/kmlab.nagaokaut.ac.jp\/index.php\/walkposture_detection\/","title":{"rendered":"\u6df1\u5c64\u5b66\u7fd2\u3092\u7528\u3044\u305f\u5358\u4e00\u52a0\u901f\u5ea6\u30bb\u30f3\u30b5\u304b\u3089\u306e\u6b69\u884c\u59ff\u52e2\u5206\u985e"},"content":{"rendered":"

Walking-posture Classification from Single-acceleration-sensor Data using Deep Learning<\/h1>\n

\u6982\u8981\uff1a\u3000\u672c\u7814\u7a76\u3067\u306f\u8170\u4ed8\u8fd1\u306b\u88c5\u7740\u3057\u305f\u5358\u4e00\u52a0\u901f\u5ea6\u30bb\u30f3\u30b5\u3067\u65e5\u5e38\u6b69\u884c\u52d5\u4f5c\u3092\u76e3\u8996\u3059\u308b\u305f\u3081\u306b\uff0c\u6df1\u5c64\u5b66\u7fd2\u3092\u5c0e\u5165\u3057\u305f\u6b69\u884c\u59ff\u52e2\u5206\u985e\u65b9\u6cd5\u3092\u691c\u8a0e\u3057\u305f\uff0e\u6df1\u5c64\u5b66\u7fd2\u3092\u9069\u7528\u3057\u305f\u65b9\u6cd5\u306fSVM\u3092\u7528\u3044\u305f\u5f93\u6765\u65b9\u6cd5\u3068\u6bd4\u8f03\u3057\uff0c\u9ad8\u3044\u5206\u985e\u7cbe\u5ea6\u304c\u5f97\u3089\u308c\uff0c\u7279\u306b\u7573\u307f\u8fbc\u307f\u5c64\u3068LSTM\u5c64\u3092\u7d44\u5408\u308f\u305b\u305f\u30cd\u30c3\u30c8\u30ef\u30fc\u30af\u69cb\u6210\u304c0.98\u3068\u6700\u3082\u9ad8\u3044\u7d50\u679c\u3067\u3042\u3063\u305f\uff0e<\/p>\n

Abstract:\u00a0 \u00a0 We described a walking-posture classification method from a single accelerator attached to a human waist using a deep learning technique. We considered deep learning architectures for a single accelerator based on previous human activity recognition studies and investigated the classification accuracy of the proposed method using the walking-posture dataset. The results demonstrate that a deep learning approach to walking-posture classification using a single accelerator is more useful than the conventional SVM approach. Additionally, we also confirmed that a hybrid network architecture with three convolutional neural layers, two pooling layers between the convolutional layers, and a long short-term memory layer achieved the best accuracy of 0.9803 compared to other network architectures. We also confirmed the deep learning approach yielded more correct classification for each walking-posture category in spite of the difficulty to detect the classification by the SVM approach.<\/p>\n

\"\" \"\"<\/p>\n

\"\"\u00a0<\/p>\n

\u7814\u7a76\u6210\u679c\uff1a<\/p>\n