CubeLearn: End-to-End Learning for Human Motion Recognition from Raw mmWave Radar Signals

Keyphrases

• Discrete Fourier Transform 100%
• Radar Signal 100%
• Edge Devices 100%
• Preprocessing Module 100%
• MmWave Radar 100%
• Mm-wave FMCW Radar 100%
• Human Motion Recognition 100%
• Computational Efficiency 50%
• Classification Accuracy 50%
• Hybrid Method 50%
• Initialization Method 50%
• Extracting Features 50%
• Overall Accuracy 50%
• Problem Recognition 50%
• Person-centered 50%
• Data Cube 50%
• Human Activity Recognition 50%
• Ablation Study 50%
• End-to-end Deep Learning 50%
• Motion Recognition 50%
• Human Gesture Recognition 50%
• Cube Slicing 50%
• Task-agnostic 50%
• Radar Recognition 50%
• Existing pipeline 50%
• Deep Neural Network Classifier 50%
• Step-back 50%

Computer Science

• Deep Neural Network 100%
• Human Motions 100%
• Fourier Transform 100%
• Preprocessing Module 100%
• Gesture Recognition 50%
• Initialization Method 50%
• Data Cube 50%
• Activity Recognition 50%
• Computational Efficiency 50%
• Recognition Problem 50%
• Classification Accuracy 50%