Human Activity Recognition
Easily identify 5 different activities with a 3D accelerometer.
Human Activity Recognition (HAR) is a time series classification task identifying the specific movement or action of a person based on sensor data. Movements can be activities performed indoors, such as walking, standing, sitting or outdoors such as driving, biking. The demo runs on a small form factor board Sensor Tile that comes along with a smartphone application connected through Bluetooth Low Energy.
Approach
- Exploits 3-axis accelerometer data
- 5 classes: stationary, walking, running, biking, driving
- Pre/Post-processing: filtering gravity, reference rotation, temporal filter
The main model is a ST Convolutional Neural Network model, but several other models are proposed within our function packs FP-AI-SENSING1 and FP-AI-MONITOR1, another CNN and a SVC model.
- 5 classes: stationary, walking, running, biking, driving
- Pre/Post-processing: filtering gravity, reference rotation, temporal filter
The main model is a ST Convolutional Neural Network model, but several other models are proposed within our function packs FP-AI-SENSING1 and FP-AI-MONITOR1, another CNN and a SVC model.
Sensor
Vision: 3D Accelerometer (reference: LSM6DSM)
Data
Data format
3D-accelerometer acquired @ 26Hz
5 activities / 185 minutes per activity
Sensor held in various places (backpack, wrist, in hand, )
3D-accelerometer acquired @ 26Hz
5 activities / 185 minutes per activity
Sensor held in various places (backpack, wrist, in hand, )
Results
Model ST Convolutional Neural Network
Input size: 24x3
Memory footprint:
12 KB Flash for weights
1.8 KBRAM for activations
Performance on STM32L476 (Low Power) @ 80 MHz
Use case: 1 classification/sec
Pre/Post-processing: 0.02 MHz
NN processing: 0.35 MHz
Power consumption (1.8 V)
- System: ~ 580 uA (with optimization BLE)
- STM32: ~ 510 uA
Input size: 24x3
Memory footprint:
12 KB Flash for weights
1.8 KBRAM for activations
Performance on STM32L476 (Low Power) @ 80 MHz
Use case: 1 classification/sec
Pre/Post-processing: 0.02 MHz
NN processing: 0.35 MHz
Power consumption (1.8 V)
- System: ~ 580 uA (with optimization BLE)
- STM32: ~ 510 uA
Resources
A free STM32Cube expansion package, X-CUBE-AI allows developers to convert pretrained AI algorithms automatically, such as neural network and machine learning models, into optimized C code for STM32.
Compatible with STM32L4 series
The STM32 family of 32-bit microcontrollers based on the Arm Cortex®-M processor is designed to offer new degrees of freedom to MCU users. It offers products combining very high performance, real-time capabilities, digital signal processing, low-power / low-voltage operation, and connectivity, while maintaining full integration and ease of development.