by Sven Mayer
The material is licensed under the Creative Commons Attribution-Share Alike 4.0 (CC BY-SA) license: https://creativecommons.org/licenses/by-sa/4.0
import os
os.environ["CUDA_VISIBLE_DEVICES"]="1"
import sys
print("Python version: ", sys.version)
import numpy as np
print("numpy version", np.__version__)
import matplotlib
import matplotlib.pyplot as plt
import matplotlib.patches as mpatches
print("matplotlib version", matplotlib.__version__)
import tensorflow as tf
print("TF:", tf.__version__)
physical_devices = tf.config.experimental.list_physical_devices('GPU')
if (len(physical_devices) > 0):
tf.config.set_visible_devices(physical_devices[0], 'GPU')
tf.config.experimental.set_memory_growth(physical_devices[0], True)
else:
print("TF: No GPU found")
import pandas as pd
from tqdm import tqdm
Python version: 3.9.1 (default, Jan 8 2021, 17:17:17) [Clang 12.0.0 (clang-1200.0.32.28)] numpy version 1.19.5 matplotlib version 3.3.3 INFO:tensorflow:Enabling eager execution INFO:tensorflow:Enabling v2 tensorshape INFO:tensorflow:Enabling resource variables INFO:tensorflow:Enabling tensor equality INFO:tensorflow:Enabling control flow v2 TF: 2.5.0-rc0 TF: No GPU found
Mi Zhang and Alexander A. Sawchuk, "USC-HAD: A Daily Activity Dataset for Ubiquitous Activity Recognition Using Wearable Sensors", ACM International Conference on Ubiquitous Computing (UbiComp) Workshop on Situation, Activity and Goal Awareness (SAGAware), Pittsburgh, Pennsylvania, USA, September 2012. DOI: https://dl.acm.org/doi/10.1145/2370216.2370438
# Download Data
import requests
url = 'http://sipi.usc.edu/had/USC-HAD.zip'
r = requests.get(url, allow_redirects=True)
open('USC-HAD.zip', 'wb').write(r.content)
# Unzip
import zipfile
with zipfile.ZipFile('USC-HAD.zip', 'r') as zip_ref:
zip_ref.extractall("./")
SAMPLING_RATE = 100 #Hz
FPS = 1/SAMPLING_RATE
44583390
file1 = open('./USC-HAD/readme.txt', 'r')
Lines = file1.readlines()
count = 0
classes = []
# Strips the newline character
for line in Lines:
if (line.strip().startswith("***")):
count += 1
if (count == 6):
if (line.strip() == "" or line.strip().startswith("***")):
pass
else:
classes.append(line.strip().split(". ")[-1])
print(classes)
['Walking Forward', 'Walking Left', 'Walking Right', 'Walking Upstairs', 'Walking Downstairs', 'Running Forward', 'Jumping Up', 'Sitting', 'Standing', 'Sleeping', 'Elevator Up', 'Elevator Down']
import scipy.io
mat = scipy.io.loadmat('./USC-HAD/Subject1/a1t1.mat')
header = ["AccX", "AccY", "AccZ", "GyroX", "GyroY", "GyroZ"]
df = pd.DataFrame(mat['sensor_readings'], columns=header)
df.head()
| AccX | AccY | AccZ | GyroX | GyroY | GyroZ | |
|---|---|---|---|---|---|---|
| 0 | 1.162780 | -0.189287 | -0.295781 | -18.781324 | 3.361872 | -16.863129 |
| 1 | 1.032296 | -0.200145 | -0.376304 | -13.544509 | 5.778864 | -14.446136 |
| 2 | 0.905245 | -0.203764 | -0.394605 | -24.823805 | 6.987360 | -21.697113 |
| 3 | 0.778195 | -0.211002 | -0.376304 | -26.032301 | 11.821344 | -21.697113 |
| 4 | 0.633975 | -0.229098 | -0.379964 | -15.961501 | 15.446833 | -7.597991 |
fig, ax = plt.subplots(2,1, figsize=(16, 5))
df[["AccX", "AccY", "AccZ"]].plot(ax=ax[0])
df[["GyroX", "GyroY", "GyroZ"]].plot(ax=ax[1])
ax[0].legend(ncol=1, loc=2)
ax[1].legend(ncol=1, loc=2)
ax[0].set_xlim(0, len(df))
ax[1].set_xlim(0, len(df))
ax[0].set_title(f"Activity {mat['activity'][0]} with sensor at {mat['sensor_location'][0]}")
plt.savefig("./figures/16_data.png", dpi=500, bbox_inches = 'tight', pad_inches = 0)
plt.show()
TIME_OF_WINDOW = 2 # seconds
WINDOW_LENGTH = int(TIME_OF_WINDOW/FPS)
inputs = tf.keras.layers.Input(shape=(WINDOW_LENGTH, 3), name="Input")
x = tf.keras.layers.SimpleRNN(32, name="RNN")(inputs)
prediction = tf.keras.layers.Dense(len(classes), activation='softmax', name="Output")(x)
model_functional = tf.keras.Model(inputs=inputs, outputs=prediction)
model_functional.summary()
Model: "model_9" _________________________________________________________________ Layer (type) Output Shape Param # ================================================================= Input (InputLayer) [(None, 200, 3)] 0 _________________________________________________________________ RNN (SimpleRNN) (None, 32) 1152 _________________________________________________________________ Output (Dense) (None, 12) 396 ================================================================= Total params: 1,548 Trainable params: 1,548 Non-trainable params: 0 _________________________________________________________________
inputs = tf.keras.layers.Input(shape=(WINDOW_LENGTH, 3), name="Input")
x = tf.keras.layers.SimpleRNN(32, name="RNN_1", return_sequences=True)(inputs)
x = tf.keras.layers.SimpleRNN(32, name="RNN_2")(x)
prediction = tf.keras.layers.Dense(len(classes), activation='softmax', name="Output")(x)
model_functional = tf.keras.Model(inputs=inputs, outputs=prediction)
model_functional.summary()
Model: "model_7" _________________________________________________________________ Layer (type) Output Shape Param # ================================================================= Input (InputLayer) [(None, 200, 3)] 0 _________________________________________________________________ RNN_1 (SimpleRNN) (None, 200, 32) 1152 _________________________________________________________________ RNN_2 (SimpleRNN) (None, 32) 2080 _________________________________________________________________ Output (Dense) (None, 12) 396 ================================================================= Total params: 3,628 Trainable params: 3,628 Non-trainable params: 0 _________________________________________________________________
inputs = tf.keras.layers.Input(shape=(WINDOW_LENGTH, 3), name="Input")
x = tf.keras.layers.LSTM(32, name="LSTM_1", return_sequences=True)(inputs)
x = tf.keras.layers.LSTM(32, name="LSTM_2")(x)
prediction = tf.keras.layers.Dense(len(classes), activation='softmax', name="Output")(x)
model_functional = tf.keras.Model(inputs=inputs, outputs=prediction)
model_functional.summary()
Model: "model_8" _________________________________________________________________ Layer (type) Output Shape Param # ================================================================= Input (InputLayer) [(None, 200, 3)] 0 _________________________________________________________________ LSTM_1 (LSTM) (None, 200, 32) 4608 _________________________________________________________________ LSTM_2 (LSTM) (None, 32) 8320 _________________________________________________________________ Output (Dense) (None, 12) 396 ================================================================= Total params: 13,324 Trainable params: 13,324 Non-trainable params: 0 _________________________________________________________________