forked from aqibsaeed/Human-Activity-Recognition-using-CNN
-
Notifications
You must be signed in to change notification settings - Fork 0
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
- Loading branch information
1 parent
53568cc
commit d685574
Showing
1 changed file
with
266 additions
and
0 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,266 @@ | ||
{ | ||
"cells": [ | ||
{ | ||
"cell_type": "code", | ||
"execution_count": 9, | ||
"metadata": { | ||
"collapsed": true | ||
}, | ||
"outputs": [], | ||
"source": [ | ||
"import pandas as pd\n", | ||
"import numpy as np\n", | ||
"import matplotlib.pyplot as plt\n", | ||
"from scipy import stats\n", | ||
"import tensorflow as tf\n", | ||
"\n", | ||
"%matplotlib inline\n", | ||
"plt.style.use('ggplot')" | ||
] | ||
}, | ||
{ | ||
"cell_type": "code", | ||
"execution_count": 10, | ||
"metadata": { | ||
"collapsed": true | ||
}, | ||
"outputs": [], | ||
"source": [ | ||
"def read_data(file_path):\n", | ||
" column_names = ['user-id','activity','timestamp', 'x-axis', 'y-axis', 'z-axis']\n", | ||
" data = pd.read_csv(file_path,header = None, names = column_names)\n", | ||
" return data\n", | ||
"\n", | ||
"def feature_normalize(dataset):\n", | ||
" mu = np.mean(dataset,axis = 0)\n", | ||
" sigma = np.std(dataset,axis = 0)\n", | ||
" return (dataset - mu)/sigma\n", | ||
" \n", | ||
"def plot_axis(ax, x, y, title):\n", | ||
" ax.plot(x, y)\n", | ||
" ax.set_title(title)\n", | ||
" ax.xaxis.set_visible(False)\n", | ||
" ax.set_ylim([min(y) - np.std(y), max(y) + np.std(y)])\n", | ||
" ax.set_xlim([min(x), max(x)])\n", | ||
" ax.grid(True)\n", | ||
" \n", | ||
"def plot_activity(activity,data):\n", | ||
" fig, (ax0, ax1, ax2) = plt.subplots(nrows = 3, figsize = (15, 10), sharex = True)\n", | ||
" plot_axis(ax0, data['timestamp'], data['x-axis'], 'x-axis')\n", | ||
" plot_axis(ax1, data['timestamp'], data['y-axis'], 'y-axis')\n", | ||
" plot_axis(ax2, data['timestamp'], data['z-axis'], 'z-axis')\n", | ||
" plt.subplots_adjust(hspace=0.2)\n", | ||
" fig.suptitle(activity)\n", | ||
" plt.subplots_adjust(top=0.90)\n", | ||
" plt.show()\n", | ||
" \n", | ||
"def windows(data, size):\n", | ||
" start = 0\n", | ||
" while start < data.count():\n", | ||
" yield start, start + size\n", | ||
" start += (size / 2)\n", | ||
"\n", | ||
"def segment_signal(data,window_size = 90):\n", | ||
" segments = np.empty((0,window_size,3))\n", | ||
" labels = np.empty((0))\n", | ||
" for (start, end) in windows(data['timestamp'], window_size):\n", | ||
" x = data[\"x-axis\"][start:end]\n", | ||
" y = data[\"y-axis\"][start:end]\n", | ||
" z = data[\"z-axis\"][start:end]\n", | ||
" if(len(dataset['timestamp'][start:end]) == window_size):\n", | ||
" segments = np.vstack([segments,np.dstack([x,y,z])])\n", | ||
" labels = np.append(labels,stats.mode(data[\"activity\"][start:end])[0][0])\n", | ||
" return segments, labels\n", | ||
"\n", | ||
"def weight_variable(shape):\n", | ||
" initial = tf.truncated_normal(shape, stddev = 0.1)\n", | ||
" return tf.Variable(initial)\n", | ||
"\n", | ||
"def bias_variable(shape):\n", | ||
" initial = tf.constant(0.0, shape = shape)\n", | ||
" return tf.Variable(initial)\n", | ||
"\n", | ||
"def depthwise_conv2d(x, W):\n", | ||
" return tf.nn.depthwise_conv2d(x,W, [1, 1, 1, 1], padding='VALID')\n", | ||
"\n", | ||
"def apply_depthwise_conv(x,kernel_size,num_channels,depth):\n", | ||
" weights = weight_variable([1, kernel_size, num_channels, depth])\n", | ||
" biases = bias_variable([depth * num_channels])\n", | ||
" return tf.nn.relu(tf.add(depthwise_conv2d(x, weights),biases))\n", | ||
" \n", | ||
"def apply_max_pool(x,kernel_size,stride_size):\n", | ||
" return tf.nn.max_pool(x, ksize=[1, 1, kernel_size, 1], \n", | ||
" strides=[1, 1, stride_size, 1], padding='VALID')" | ||
] | ||
}, | ||
{ | ||
"cell_type": "code", | ||
"execution_count": 11, | ||
"metadata": { | ||
"collapsed": false | ||
}, | ||
"outputs": [], | ||
"source": [ | ||
"dataset = read_data('actitracker_raw.txt')\n", | ||
"dataset['x-axis'] = feature_normalize(dataset['x-axis'])\n", | ||
"dataset['y-axis'] = feature_normalize(dataset['y-axis'])\n", | ||
"dataset['z-axis'] = feature_normalize(dataset['z-axis'])" | ||
] | ||
}, | ||
{ | ||
"cell_type": "code", | ||
"execution_count": null, | ||
"metadata": { | ||
"collapsed": false | ||
}, | ||
"outputs": [], | ||
"source": [ | ||
"for activity in np.unique(dataset[\"activity\"]):\n", | ||
" subset = dataset[dataset[\"activity\"] == activity][:180]\n", | ||
" plot_activity(activity,subset)" | ||
] | ||
}, | ||
{ | ||
"cell_type": "code", | ||
"execution_count": null, | ||
"metadata": { | ||
"collapsed": false | ||
}, | ||
"outputs": [], | ||
"source": [ | ||
"segments, labels = segment_signal(dataset)\n", | ||
"labels = np.asarray(pd.get_dummies(labels), dtype = np.int8)\n", | ||
"reshaped_segments = segments.reshape(len(segments), 1,90, 3)" | ||
] | ||
}, | ||
{ | ||
"cell_type": "code", | ||
"execution_count": null, | ||
"metadata": { | ||
"collapsed": true | ||
}, | ||
"outputs": [], | ||
"source": [ | ||
"train_test_split = np.random.rand(len(reshaped_segments)) < 0.70\n", | ||
"train_x = reshaped_segments[train_test_split]\n", | ||
"train_y = labels[train_test_split]\n", | ||
"test_x = reshaped_segments[~train_test_split]\n", | ||
"test_y = labels[~train_test_split]" | ||
] | ||
}, | ||
{ | ||
"cell_type": "code", | ||
"execution_count": 12, | ||
"metadata": { | ||
"collapsed": false | ||
}, | ||
"outputs": [], | ||
"source": [ | ||
"input_height = 1\n", | ||
"input_width = 90\n", | ||
"num_labels = 6\n", | ||
"num_channels = 3\n", | ||
"\n", | ||
"batch_size = 10\n", | ||
"kernel_size = 60\n", | ||
"depth = 60\n", | ||
"num_hidden = 1000\n", | ||
"\n", | ||
"learning_rate = 0.0001\n", | ||
"training_epochs = 8\n", | ||
"\n", | ||
"total_batchs = reshaped_segments.shape[0] // batch_size" | ||
] | ||
}, | ||
{ | ||
"cell_type": "code", | ||
"execution_count": 14, | ||
"metadata": { | ||
"collapsed": false | ||
}, | ||
"outputs": [], | ||
"source": [ | ||
"X = tf.placeholder(tf.float32, shape=[None,input_height,input_width,num_channels])\n", | ||
"Y = tf.placeholder(tf.float32, shape=[None,num_labels])\n", | ||
"\n", | ||
"c = apply_depthwise_conv(X,kernel_size,num_channels,depth)\n", | ||
"p = apply_max_pool(c,20,2)\n", | ||
"c = apply_depthwise_conv(p,6,depth*num_channels,depth//10)\n", | ||
"\n", | ||
"shape = c.get_shape().as_list()\n", | ||
"c_flat = tf.reshape(c, [-1, shape[1] * shape[2] * shape[3]])\n", | ||
"\n", | ||
"f_weights_l1 = weight_variable([shape[1] * shape[2] * depth * num_channels * (depth//10), num_hidden])\n", | ||
"f_biases_l1 = bias_variable([num_hidden])\n", | ||
"f = tf.nn.tanh(tf.add(tf.matmul(c_flat, f_weights_l1),f_biases_l1))\n", | ||
"\n", | ||
"out_weights = weight_variable([num_hidden, num_labels])\n", | ||
"out_biases = bias_variable([num_labels])\n", | ||
"y_ = tf.nn.softmax(tf.matmul(f, out_weights) + out_biases)" | ||
] | ||
}, | ||
{ | ||
"cell_type": "code", | ||
"execution_count": null, | ||
"metadata": { | ||
"collapsed": true | ||
}, | ||
"outputs": [], | ||
"source": [ | ||
"loss = -tf.reduce_sum(Y * tf.log(y_))\n", | ||
"optimizer = tf.train.GradientDescentOptimizer(learning_rate = learning_rate).minimize(loss)\n", | ||
"\n", | ||
"correct_prediction = tf.equal(tf.argmax(y_,1), tf.argmax(Y,1))\n", | ||
"accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))" | ||
] | ||
}, | ||
{ | ||
"cell_type": "code", | ||
"execution_count": null, | ||
"metadata": { | ||
"collapsed": false, | ||
"scrolled": true | ||
}, | ||
"outputs": [], | ||
"source": [ | ||
"cost_history = np.empty(shape=[1],dtype=float)\n", | ||
"\n", | ||
"with tf.Session() as session:\n", | ||
" tf.initialize_all_variables().run()\n", | ||
" for epoch in range(training_epochs):\n", | ||
" for b in range(total_batchs): \n", | ||
" offset = (b * batch_size) % (train_y.shape[0] - batch_size)\n", | ||
" batch_x = train_x[offset:(offset + batch_size), :, :, :]\n", | ||
" batch_y = train_y[offset:(offset + batch_size), :]\n", | ||
" _, c = session.run([optimizer, loss],feed_dict={X: batch_x, Y : batch_y})\n", | ||
" cost_history = np.append(cost_history,c)\n", | ||
" print \"Epoch: \",epoch,\" Training Loss: \",c,\" Training Accuracy: \",\n", | ||
" session.run(accuracy, feed_dict={X: train_x, Y: train_y})\n", | ||
" \n", | ||
" print \"Testing Accuracy:\", session.run(accuracy, feed_dict={X: test_x, Y: test_y})" | ||
] | ||
} | ||
], | ||
"metadata": { | ||
"anaconda-cloud": {}, | ||
"kernelspec": { | ||
"display_name": "Python [conda root]", | ||
"language": "python", | ||
"name": "conda-root-py" | ||
}, | ||
"language_info": { | ||
"codemirror_mode": { | ||
"name": "ipython", | ||
"version": 2 | ||
}, | ||
"file_extension": ".py", | ||
"mimetype": "text/x-python", | ||
"name": "python", | ||
"nbconvert_exporter": "python", | ||
"pygments_lexer": "ipython2", | ||
"version": "2.7.12" | ||
} | ||
}, | ||
"nbformat": 4, | ||
"nbformat_minor": 1 | ||
} |