Robust Unsupervised Factory Activity Recognition with Body-worn Accelerometer Using Temporal ...
summary
Unsupervised system segments repetitive assembly cycles from body-worn accelerometer data by tracking period motifs that occur once per cycle and action motifs that repeat within cycles, using autocorrelation-based motif matching and relative temporal consistency to detect outlier activities. It replaces slow DTW with feature vectors and Euclidean distance for action matching, runs much faster, and on six real factory datasets reports higher average accuracy than the prior method while reducing computation time.
AI-written summary · about
related talks
28mDeep Learning-Based Human Activity Recognition for Physical Skills...
27mDissecting ActivityRecognition with Multi-Sensor Simulation and Its Practical Deployment
06mUnsupervised Human Activity Representation Learning with Multi-task Deep Clustering
06mApproaching the Real-World: Supporting Activity Recognition Training with Virtual IMU Data
22mSeeing, Sensing and Recognizing Laban Movement Qualities
06mDeriving Effective Human Activity Recognition Systems through Objective Task Complexity Assessment
11mBootstrapping User-Defined Body Tapping Recognition with Offline-Learned Probabilistic ...
22mRunning, walking, sitting or biking? - Motion prediction with acceleration and rotation