Wearable Data Analysis

Apple Watch data engine framework, dataset, and algorithms for human activity recognition.

Table of Contents

• Get Started
• Run Data Engine
• Visualized Data Samples
• Human Activity Recognition with Convolutional Neural Network
• Requirements
• References

Get Started

$ git clone [email protected]:blakete/Wearable-Data-Analysis.git

Run Data Engine

$ python3 training_data_flow_from_directory.py /path/to/dataset
processing raw dataset
.
. (processes all CSVs in target directories' sub-directories)
.
Successfully processed 100.0% of the dataset
Total samples: 61990.0
Failed samples: 0.0
Successful samples: 61990.0

Classes: ['drive', 'dustbin', 'lay', 'run', 'sit', 'skate', 'stair', 'stand', 'walk'] 
Samples: [15141.  1004. 15477.  8261. 11659.   185.   532.  3877.  6854.]

Visualized Data Samples

Human Activity Recognition

Achieved 95.9% top-1 and 100% top-3 accuracy on test dataset

Convolutional Neural Network (CNN) Architecture:

Model: "sequential"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
conv2d (Conv2D)              (None, 43, 1, 32)         320       
_________________________________________________________________
conv2d_1 (Conv2D)            (None, 42, 1, 64)         4160      
_________________________________________________________________
batch_normalization (BatchNo (None, 42, 1, 64)         256       
_________________________________________________________________
flatten (Flatten)            (None, 2688)              0         
_________________________________________________________________
dense (Dense)                (None, 64)                172096    
_________________________________________________________________
batch_normalization_1 (Batch (None, 64)                256       
_________________________________________________________________
activation (Activation)      (None, 64)                0         
_________________________________________________________________
dense_1 (Dense)              (None, 32)                2080      
_________________________________________________________________
batch_normalization_2 (Batch (None, 32)                128       
_________________________________________________________________
activation_1 (Activation)    (None, 32)                0         
_________________________________________________________________
dense_2 (Dense)              (None, 7)                 231       
=================================================================
Total params: 179,527
Trainable params: 179,207
Non-trainable params: 320

Example CNN Results:

Epoch 150/150
62273/62273 [==============================] - 12s 196us/sample - val_loss: 0.1124 - val_accuracy: 0.9593
loss: 0.11244082237378554
accuracy: 0.95932424

Confusion matrix:

Requirements

• >= Apple A12 or >= iPhone XR
• >= Apple Watch Series 3
• Wearable CoPilot App (coming soon!)

References

Cole, C.A., Janos, B., Anshari, D., Thrasher, J.F., Strayer, S.M., & Valafar, H. (2020). Recognition of Smoking Gesture Using Smart Watch Technology. ArXiv, abs/2003.02735.

Twomey, N.; Diethe, T.; Fafoutis, X.; Elsts, A.; McConville, R.; Flach, P.; Craddock, I. A Comprehensive Study of Activity Recognition Using Accelerometers. Informatics 2018, 5, 27.