An activity recognition method based on deep learning and independent of device orientation and placement was proposed, to address the problem that traditional acceleration activity recognition methods usually depend on the fixed device orientations and placements. Based on a layerwise unsupervised learning and a global supervised finetuning, deep neural networks were constructed by stacked autoencoder. Then, the deep features of the proposed method were efficiently and effectively extracted from original acceleration data. Finally, a cross orientation and placement evaluation strategy was presented to recognize the activities under different device orientations and placements. Experimental results show that the proposed method can extract discriminative deep features from original data and achieve better performance than other methods under the condition of uncontrolled acceleration device orientation and placement.
SHEN Yan bin, CHEN Ling, GUO Hao dong, CHEN Gen cai. Deep learning based activity recognition independent of device orientation and placement. JOURNAL OF ZHEJIANG UNIVERSITY (ENGINEERING SCIENCE), 2016, 50(6): 1141-1148.
［1］ AMFT O, STGER M, LUKOWICZ P, et al. Analysis of chewing sounds for dietary monitoring ［C］ ∥ Proceedings of International Conference on Ubiquitous Computing. Tokyo: Springer, 2005: 56-72.
［2］ ALBINALI F, GOODWIN M S, INTILLE S S. Recognizing stereotypical motor movements in the laboratory and classroom: a case study with children on the autism spectrum ［C］ ∥ Proceedings of International Conference on Ubiquitous Computing. Orlando: ACM, 2009: 71-80.
［3］ LADHA C, HAMMERLA N Y, OLIVIER P, et al. ClimbAX: skill assessment for climbing enthusiasts ［C］ ∥ Proceedings of International Joint Conference on Pervasive and Ubiquitous Computing. Zurich: ACM, 2013: 235-244.
［4］ ATALLAH L, YANG G Z. The use of pervasive sensing for behaviour profiling: a survey ［J］. Pervasive and Mobile Computing, 2009, 5(5): 447-464.
［5］ DEVAUL R W, DUNN S. Realtime motion classification for wearable computing applications ［R］. Cambridge: MIT Media Laboratory, 2001.
［6］ LEE S W, MASE K. Activity and location recognition using wearable sensors ［J］. Pervasive Computing, 2002, 1(3): 24-32.
［7］ BAO L, INTILLE S S. Activity recognition from userannotated acceleration data ［C］ ∥ Proceedings of International Conference on Pervasive Computing. Vienna: Springer, 2004: 117.
［8］ KWAPISZ J R, WEISS G M, MOORE S A. Activity recognition using cell phone accelerometers ［J］. ACM SIGKDD Explorations Newsletter, 2011, 12(2): 74-82.
［9］ RAVI N, DANDEKAR N, MYSORE P, et al. Activity recognition from accelerometer data ［C］ ∥ Proceedings of Conference on Innovative Applications of Artificial Intelligence. Pittsburgh: AAAI, 2005: 1541-1546.
［10］ OLGUIN D O, PENTLAND A S. Human activity recognition: Accuracy across common locations for wearable sensors ［C］ ∥ Proceedings of International Symposium on Wearable Computers (Student Colloquium). Montreux: IEEE, 2006: 11-13.
［11］ BAYATI H, DEL R MILLN J, CHAVARRIAGA R. Unsupervised adaptation to onbody sensor displacement in accelerationbased activity recognition ［C］ ∥ Proceedings of International Symposium on Wearable Computers. San Francisco: IEEE, 2011: 71-78.
［12］ CHAVARRIAGA R, BAYATI H, MILLN J D. Unsupervised adaptation for accelerationbased activity recognition: robustness to sensor displacement androtation ［J］. Personal and Ubiquitous Computing, 2013, 17(3): 479-490.
［13］ 侯仓健,陈岭,吕明琪,等.基于加速度传感器的放置方式和位置无关运动识别［J］.计算机科学,2014, 41(10): 7679, 94.
HOU Cangjian, CHEN Ling, LV Mingqi, et al. Accelerationbased activity recognition independent of device orientation and placement ［J］. Computer Science, 2014, 41(10): 7679, 94.
［14］ LESTER J, CHOUDHURY T, BORRIELLO G. A practical approach to recognizing physical activities ［C］ ∥Proceedings of International Conference on PERVASIVE. Dublin: Springer, 2006: 116.
［15］ PLTZ T, HAMMERLA N Y, OLIVIER P. Feature learning for activity recognition in ubiquitous computing ［C］ ∥ Proceedings of International Joint Conference on Artificial Intelligence. Barcelona: AAAI, 2011: 1729-1734.
［16］ VOLLMER C, GROSS H M, EGGERT J P. Learning Features for Activity Recognition with ShiftInvariant Sparse Coding ［C］ ∥ Proceedings of International Conference on Artificial Neural Networks. Sofia: Springer, 2013: 367-374.
［17］ ZENG M, NGUYEN L T, YU B, et al. Convolutional neural networks for human activity recognition using mobile sensors ［C］ ∥ Proceedings of Sixth International Conference on Mobile Computing, Applications and Services. Austin: IEEE, 2014: 197-205.
［18］ GUO H, CHEN L, CHEN G, LV M. Smartphonebased activity recognition independent of device orientation and placement ［J］. International Journal of Communication Systems, 2015.
［19］ MARTINEZ H P, BENGIO Y, YANNAKAKIS G N. Learning deep physiological models of affect ［J］. Computational Intelligence Magazine, IEEE, 2013, 8(2): 2033.
［20］ PALM R B. Prediction as a candidate for learning deep hierarchical models of data ［J］. Technical University of Denmark, Palm, 2012, 25.
YANG Bing, WANG Xiao-hua, YANG Xin, HUANG Xiao-xi. Face recognition method based on HOG pyramid[J]. JOURNAL OF ZHEJIANG UNIVERSITY (ENGINEERING SCIENCE), 2014, 48(9): 1564-1569.
