Lab Home | Dtaset Home               Chinese | English


A brief introduction of SCUT-NAA

  In order to facilitate the research of accelerometer-based human activity recognition, we build a naturalistic 3D acceleration-based activity dataset, SCUT-NAA dataset. It’s going to provide researchers in the field of acceleration-based activity recognition with a naturalistic activity dataset as training and testing samples and it’s also a standard dataset for comparing and evaluating performance of different algorithms.

  SCUT-NAA dataset is the first public available 3D acceleration-based activity dataset that contains 1278 samples of 44 individuals (34 males and 10 females) which collected in naturalistic settings with only one tri-axial accelerometer located in waist belt, pants pocket, and cloth pocket respectively. Each sampling person is asked to perform ten activities.

  For more details of SCUT-NAA dataset, please refer our paper: Yang Xue, Lianwen, Jin. “A Naturalistic 3D Acceleration-based Activity Dataset & Benchmark Evaluations”, IEEE International Conference on Systems, Man, and Cybernetics,pp.4081-4085, 2010.