我一直在关注以下tensorflow教程，在python中进行培训并在android中进行部署。
培训代码：
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from scipy import stats
import tensorflow as tf
from tensorflow.python.framework.graph_util import convert_variables_to_constants
# Just disables the warning, doesn't enable AVX/FMA
import os
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'
#matplotlib inline
plt.style.use('ggplot')
def read_data(file_path):
column_names = ['user-id', 'activity', 'timestamp', 'x-axis', 'y-axis', 'z-axis']
data = pd.read_csv(file_path, header=None, names=column_names)
return data
def feature_normalize(dataset):
mu = np.mean(dataset, axis=0)
sigma = np.std(dataset, axis=0)
return (dataset - mu) / sigma
def plot_axis(ax, x, y, title):
ax.plot(x, y)
ax.set_title(title)
ax.xaxis.set_visible(False)
ax.set_ylim([min(y) - np.std(y), max(y) + np.std(y)])
ax.set_xlim([min(x), max(x)])
ax.grid(True)
def plot_activity(activity, data):
fig, (ax0, ax1, ax2) = plt.subplots(nrows=3, figsize=(15, 10), sharex=True)
plot_axis(ax0, data['timestamp'], data['x-axis'], 'x-axis')
plot_axis(ax1, data['timestamp'], data['y-axis'], 'y-axis')
plot_axis(ax2, data['timestamp'], data['z-axis'], 'z-axis')
plt.subplots_adjust(hspace=0.2)
fig.suptitle(activity)
plt.subplots_adjust(top=0.90)
plt.show()
def windows(data, size):
start = 0
while start < data.count():
yield int(start), int(start + size)
start += (size / 2)
def segment_signal(data, window_size=90):
segments = np.empty((0, window_size, 3))
labels = np.empty((0))
for (start, end) in windows(data['timestamp'], window_size):
x = data["x-axis"][start:end]
y = data["y-axis"][start:end]
z = data["z-axis"][start:end]
if (len(dataset['timestamp'][start:end]) == window_size):
segments = np.vstack([segments, np.dstack([x, y, z])])
labels = np.append(labels, stats.mode(data["activity"][start:end])[0][0])
return segments, labels
def weight_variable(shape):
initial = tf.truncated_normal(shape, stddev=0.1)
return tf.Variable(initial)
def bias_variable(shape):
initial = tf.constant(0.0, shape=shape)
return tf.Variable(initial)
def depthwise_conv2d(x, W):
return tf.nn.depthwise_conv2d(x, W, [1, 1, 1, 1], padding='VALID')
def apply_depthwise_conv(x, kernel_size, num_channels, depth):
weights = weight_variable([1, kernel_size, num_channels, depth])
biases = bias_variable([depth * num_channels])
return tf.nn.relu(tf.add(depthwise_conv2d(x, weights), biases))
def apply_max_pool(x, kernel_size, stride_size):
return tf.nn.max_pool(x, ksize=[1, 1, kernel_size, 1],
strides=[1, 1, stride_size, 1], padding='VALID')
dataset = read_data('WISDM_ar_v1.1_raw_copy.txt')
dataset.dropna(axis=0, how='any', inplace= True)
dataset.drop_duplicates(['user-id','activity','timestamp', 'x-axis', 'y-axis', 'z-axis'], keep= 'first', inplace= True)
dataset['x-axis'] = feature_normalize(dataset['x-axis'])
dataset['y-axis'] = feature_normalize(dataset['y-axis'])
dataset['z-axis'] = feature_normalize(dataset['z-axis'])
# for activity in np.unique(dataset["activity"]):
# subset = dataset[dataset["activity"] == activity][:180]
# plot_activity(activity,subset)
segments, labels = segment_signal(dataset)
labels = np.asarray(pd.get_dummies(labels), dtype=np.int8)
reshaped_segments = segments.reshape(len(segments), 1, 90, 3)
train_test_split = np.random.rand(len(reshaped_segments)) < 0.70
train_x = reshaped_segments[train_test_split]
train_y = labels[train_test_split]
test_x = reshaped_segments[~train_test_split]
test_y = labels[~train_test_split]
input_height = 1 # 1-Dimensional convulotion
input_width = 90 #window
num_labels = 6 #output labels
num_channels = 3 #input columns
batch_size = 10
kernel_size = 60
depth = 60
num_hidden = 1000
learning_rate = 0.0001
training_epochs = 1#8
total_batches = train_x.shape[0] # batch_size
X = tf.placeholder(tf.float32, shape=[None,input_height,input_width,num_channels])
Y = tf.placeholder(tf.float32, shape=[None,num_labels])
c = apply_depthwise_conv(X,kernel_size,num_channels,depth)
p = apply_max_pool(c,20,2)
c = apply_depthwise_conv(p,6,depth*num_channels,depth//10)
shape = c.get_shape().as_list()
c_flat = tf.reshape(c, [-1, shape[1] * shape[2] * shape[3]])
f_weights_l1 = weight_variable([shape[1] * shape[2] * depth * num_channels * (depth//10), num_hidden])
f_biases_l1 = bias_variable([num_hidden])
f = tf.nn.tanh(tf.add(tf.matmul(c_flat, f_weights_l1),f_biases_l1))
out_weights = weight_variable([num_hidden, num_labels])
out_biases = bias_variable([num_labels])
y_ = tf.nn.softmax(tf.matmul(f, out_weights) + out_biases,name="y_")
loss = -tf.reduce_sum(Y * tf.log(y_))
optimizer = tf.train.GradientDescentOptimizer(learning_rate = learning_rate).minimize(loss)
correct_prediction = tf.equal(tf.argmax(y_,1), tf.argmax(Y,1)) #difference between correct output and expected output
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
cost_history = np.empty(shape=[1], dtype=float)
with tf.Session() as session:
tf.global_variables_initializer().run()
for epoch in range(training_epochs):
for b in range(total_batches):
offset = (b * batch_size) % (train_y.shape[0] - batch_size)
batch_x = train_x[offset:(offset + batch_size), :, :, :]
batch_y = train_y[offset:(offset + batch_size), :]
_, c = session.run([optimizer, loss], feed_dict={X: batch_x, Y: batch_y})
cost_history = np.append(cost_history, c)
print "Epoch: ", epoch, " Training Loss: ", c, " Training Accuracy: ",
session.run(accuracy, feed_dict={X: train_x, Y: train_y})
print "Testing Accuracy:", session.run(accuracy, feed_dict={X: test_x, Y: test_y})
saver = tf.train.Saver()
tf.train.write_graph(session.graph_def, '.', './model/har.pbtxt')
saver.save(session, save_path="./model/har.ckpt")
# File_writer = tf.summary.FileWriter('./graph', session.graph)
冻结模型：
from tensorflow.python.tools import freeze_graph
from tensorflow.python.tools import optimize_for_inference_lib
import tensorflow as tf
freeze_graph.freeze_graph(input_graph = "./model/har.pbtxt", input_saver = "",
input_binary = False, input_checkpoint = "./model/har.ckpt", output_node_names = "y_",
restore_op_name = "save/restore_all", filename_tensor_name = "save/Const:0",
output_graph = "./model/frozen_har.pb", clear_devices = True, initializer_nodes = "")
input_graph_def = tf.GraphDef()
with tf.gfile.Open("./model/frozen_har.pb", "r") as f:
data = f.read()
input_graph_def.ParseFromString(data)
output_graph_def = optimize_for_inference_lib.optimize_for_inference(
input_graph_def,
["input"],
["y_"],
tf.float32.as_datatype_enum)
f = tf.gfile.FastGFile("./model/optimized_har.pb", "w")
f.write(output_graph_def.SerializeToString())
当我尝试冻结模型时出现错误：
2018-04-05 10:00:50.604122: I tensorflow/core/platform/cpu_feature_guard.cc:140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
Converted 8 variables to const ops.
Traceback (most recent call last):
File "/home/aawesh/PycharmProjects/Human-Activity-Recognition-using-CNN/Freeze.py", line 19, in <module>
tf.float32.as_datatype_enum)
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/tools/optimize_for_inference_lib.py", line 109, in optimize_for_inference
placeholder_type_enum)
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/tools/strip_unused_lib.py", line 83, in strip_unused
raise KeyError("The following input nodes were not found: %s\n" % not_found)
KeyError: "The following input nodes were not found: set(['input'])\n"
Process finished with exit code 1
Android部分：
MainActivity.java
package io.github.aqibsaeed.activityrecognition;
import android.content.Context;
import android.hardware.Sensor;
import android.hardware.SensorEvent;
import android.hardware.SensorEventListener;
import android.hardware.SensorManager;
import android.os.Bundle;
import android.support.v7.app.AppCompatActivity;
import android.widget.TextView;
import java.math.BigDecimal;
import java.util.ArrayList;
import java.util.List;
public class MainActivity extends AppCompatActivity implements SensorEventListener {
private final int N_SAMPLES = 90;
private static List<Float> x;
private static List<Float> y;
private static List<Float> z;
private static List<Float> input_signal;
private SensorManager mSensorManager;
private Sensor mAccelerometer;
private ActivityInference activityInference;
private TextView downstairsTextView;
private TextView joggingTextView;
private TextView sittingTextView;
private TextView standingTextView;
private TextView upstairsTextView;
private TextView walkingTextView;
@Override
protected void onCreate(Bundle savedInstanceState) {
super.onCreate(savedInstanceState);
setContentView(R.layout.activity_main);
x = new ArrayList<Float>();
y = new ArrayList<Float>();
z = new ArrayList<Float>();
input_signal = new ArrayList<Float>();
downstairsTextView = (TextView)findViewById(R.id.downstairs_prob);
joggingTextView = (TextView)findViewById(R.id.jogging_prob);
sittingTextView = (TextView)findViewById(R.id.sitting_prob);
standingTextView = (TextView)findViewById(R.id.standing_prob);
upstairsTextView = (TextView)findViewById(R.id.upstairs_prob);
walkingTextView = (TextView)findViewById(R.id.walking_prob);
mSensorManager = (SensorManager) getSystemService(Context.SENSOR_SERVICE);
mAccelerometer = mSensorManager.getDefaultSensor(Sensor.TYPE_ACCELEROMETER);
mSensorManager.registerListener(this, mAccelerometer , SensorManager.SENSOR_DELAY_FASTEST);
activityInference = new ActivityInference(getApplicationContext());
}
protected void onPause() {
super.onPause();
mSensorManager.unregisterListener(this);
}
protected void onResume() {
super.onResume();
mSensorManager.registerListener(this, mAccelerometer, SensorManager.SENSOR_DELAY_FASTEST);
}
@Override
public void onSensorChanged(SensorEvent event) {
activityPrediction();
x.add(event.values[0]);
y.add(event.values[1]);
z.add(event.values[2]);
}
@Override
public void onAccuracyChanged(Sensor sensor, int i) {
}
private void activityPrediction()
{
if(x.size() == N_SAMPLES && y.size() == N_SAMPLES && z.size() == N_SAMPLES) {
// Mean normalize the signal
normalize();
// Copy all x,y and z values to one array of shape N_SAMPLES*3
input_signal.addAll(x); input_signal.addAll(y); input_signal.addAll(z);
// Perform inference using Tensorflow
float[] results = activityInference.getActivityProb(toFloatArray(input_signal));
downstairsTextView.setText(Float.toString(round(results[0],2)));
joggingTextView.setText(Float.toString(round(results[1],2)));
sittingTextView.setText(Float.toString(round(results[2],2)));
standingTextView.setText(Float.toString(round(results[3],2)));
upstairsTextView.setText(Float.toString(round(results[4],2)));
walkingTextView.setText(Float.toString(round(results[5],2)));
// Clear all the values
x.clear(); y.clear(); z.clear(); input_signal.clear();
}
}
private float[] toFloatArray(List<Float> list)
{
int i = 0;
float[] array = new float[list.size()];
for (Float f : list) {
array[i++] = (f != null ? f : Float.NaN);
}
return array;
}
private void normalize()
{
float x_m = 0.662868f; float y_m = 7.255639f; float z_m = 0.411062f;
float x_s = 6.849058f; float y_s = 6.746204f; float z_s = 4.754109f;
for(int i = 0; i < N_SAMPLES; i++){
x.set(i,((x.get(i) - x_m)/x_s));
y.set(i,((y.get(i) - y_m)/y_s));
z.set(i,((z.get(i) - z_m)/z_s));
}
}
public static float round(float d, int decimalPlace) {
BigDecimal bd = new BigDecimal(Float.toString(d));
bd = bd.setScale(decimalPlace, BigDecimal.ROUND_HALF_UP);
return bd.floatValue();
}
}
ActivityInference.java
package io.github.aqibsaeed.activityrecognition;
import android.content.Context;
import android.content.res.AssetManager;
import org.tensorflow.contrib.android.TensorFlowInferenceInterface;
public class ActivityInference {
// static {
// System.loadLibrary("tensorflow_inference");
// }
private static ActivityInference activityInferenceInstance;
private TensorFlowInferenceInterface inferenceInterface;
// private static final String MODEL_FILE = "file:///android_asset/optimized_har.pb";
private static final String MODEL_FILE = "file:///android_asset/optimized_har_1.pb";
private static final String INPUT_NODE = "input";
private static final String[] OUTPUT_NODES = {"y_"};
private static final String OUTPUT_NODE = "y_";
private static final long[] INPUT_SIZE = {1,270};
private static final int OUTPUT_SIZE = 6;
private static AssetManager assetManager;
public static ActivityInference getInstance(final Context context)
{
if (activityInferenceInstance == null)
{
activityInferenceInstance = new ActivityInference(context);
}
return activityInferenceInstance;
}
public ActivityInference(final Context context) {
this.assetManager = context.getAssets();
inferenceInterface = new TensorFlowInferenceInterface(assetManager, MODEL_FILE);
}
public float[] getActivityProb(float[] input_signal)
{
float[] result = new float[OUTPUT_SIZE];
inferenceInterface.feed(INPUT_NODE,input_signal,INPUT_SIZE);
inferenceInterface.run(OUTPUT_NODES);
inferenceInterface.fetch(OUTPUT_NODE,result);
//Downstairs Jogging Sitting Standing Upstairs Walking
return result;
}
}
Implementation of CNN for Human Activity Recognition
Deploying the trained model in android
我的问题是：
在部署时，它使用名为＆＃39; input＆＃39;的节点。但是我们在训练时没有这样的张量名称。冻结图形时会出现问题。我该如何解决？
在训练模型时，它使用形状4维的输入张量。但是，在android中我们只提供一个列表。它是如何工作的？
请帮帮我。我被困在这一段很长一段时间了。
Similar question has been asked here 但它不是重复的。我也没有在那里找到答案。
我的尝试：我以某种方式设法释放模型并在android上运行它。我收到了以下错误。我仍然需要帮助来冻结模型：
04-05 10:16:43.988 13148-13148/tictactoe.com.tictacoe E/AndroidRuntime: FATAL EXCEPTION: main
Process: tictactoe.com.tictacoe, PID: 13148
java.lang.IllegalArgumentException: input must be 4-dimensional[1,270]
[[Node: depthwise = DepthwiseConv2dNative[T=DT_FLOAT, data_format="NHWC", dilations=[1, 1, 1, 1], padding="VALID", strides=[1, 1, 1, 1], _device="/job:localhost/replica:0/task:0/device:CPU:0"](_arg_input_0_0, Variable)]]
at org.tensorflow.Session.run(Native Method)
at org.tensorflow.Session.access$100(Session.java:48)
at org.tensorflow.Session$Runner.runHelper(Session.java:298)
at org.tensorflow.Session$Runner.run(Session.java:248)
at org.tensorflow.contrib.android.TensorFlowInferenceInterface.run(TensorFlowInferenceInterface.java:230)
at org.tensorflow.contrib.android.TensorFlowInferenceInterface.run(TensorFlowInferenceInterface.java:197)
at org.tensorflow.contrib.android.TensorFlowInferenceInterface.run(TensorFlowInferenceInterface.java:187)
at io.github.aqibsaeed.activityrecognition.ActivityInference.getActivityProb(ActivityInference.java:43)
at io.github.aqibsaeed.activityrecognition.MainActivity.activityPrediction(MainActivity.java:89)
at io.github.aqibsaeed.activityrecognition.MainActivity.onSensorChanged(MainActivity.java:68)
at android.hardware.SystemSensorManager$SensorEventQueue.dispatchSensorEvent(SystemSensorManager.java:503)
at android.os.MessageQueue.nativePollOnce(Native Method)
at android.os.MessageQueue.next(MessageQueue.java:143)
at android.os.Looper.loop(Looper.java:130)
at android.app.ActivityThread.main(ActivityThread.java:6895)
at java.lang.reflect.Method.invoke(Native Method)
at java.lang.reflect.Method.invoke(Method.java:372)
at com.android.internal.os.ZygoteInit$MethodAndArgsCaller.run(ZygoteInit.java:1404)
at com.android.internal.os.ZygoteInit.main(ZygoteInit.java:1199)