Adding model design and training code

ANN_validation.py

@@ -0,0 +1,33 @@
+import model_audio_ANN
+import pickle as cPickle
+import tensorflow as tf
+import tflearn.helpers.evaluator
+
+
+print("Load evaluation dataset")
+pickle_path = './pickle_data/'
+
+
+# f = open(pickle_path+'train_labels_22k_org.pickle', 'wb')
+# f = open(pickle_path+'valid_labels_22k_org.pickle', 'wb')
+# with open(pickle_path+"test_data_22k_org.pickle", "rb") as input_file:
+with open(pickle_path + 'valid_data_22k_org.pickle', 'rb') as input_file:
+    x_test = cPickle.load(input_file)
+
+# with open(pickle_path+"test_labels_22k_org.pickle", "rb") as input_file:
+with open(pickle_path + "valid_labels_22k_org.pickle", "rb") as input_file:
+    y_test = cPickle.load(input_file)
+
+
+tf.reset_default_graph()
+
+# ann_model_dir = './model/ANN/'
+ann_model_dir = './audio_model/ANN/'
+# ann_model_dir = './model/ANN_new/Bee_audio_ANN.tfl'
+
+ann_model = model_audio_ANN.build_tflearn_ann(x_test.shape[1])
+ann_model.load(ann_model_dir+'Bee_audio_ANN.tfl', weights_only=True, create_new_session = False)
+validation_acc = ann_model.evaluate(x_test, y_test, batch_size=64)
+print(validation_acc)
+
+# Best Acc: 0.9365351630524951

CNN_validation.py

@@ -0,0 +1,27 @@
+import model_audio_CNN
+import pickle as cPickle
+import tensorflow as tf
+import tflearn.helpers.evaluator
+
+
+print("Load evaluation dataset")
+pickle_path = './pickle_data/'
+
+with open(pickle_path + 'valid_data_22k_org.pickle', 'rb') as input_file:
+# with open(pickle_path+"test_data_22k_org.pickle", "rb") as input_file:
+    x_test = cPickle.load(input_file)
+
+with open(pickle_path + "valid_labels_22k_org.pickle", "rb") as input_file:
+# with open(pickle_path+"test_labels_22k_org.pickle", "rb") as input_file:
+    y_test = cPickle.load(input_file)
+
+
+tf.reset_default_graph()
+
+cnn_model_dir = './audio_model/CNN/Bee_audio_CNN.tfl'
+cnn_model = model_audio_CNN.build_tflearn_cnn(x_test.shape[1])
+cnn_model.load(cnn_model_dir, weights_only=True)
+validation_acc = cnn_model.evaluate(x_test, y_test, batch_size=32)
+print(validation_acc)
+
+# Best Acc: 0.9948542024013722

model_audio_ANN.py

@@ -0,0 +1,71 @@
+import numpy as np
+import tensorflow as tf
+import tflearn
+import tflearn.layers.merge_ops
+from tflearn.data_utils import shuffle, to_categorical
+from tflearn.layers.core import input_data, dropout, fully_connected, flatten, reshape
+from tflearn.layers.conv import conv_2d, max_pool_2d
+from tflearn.layers.conv import conv_1d, max_pool_1d, avg_pool_1d
+from tflearn.layers.estimator import regression
+from tflearn.layers.normalization import batch_normalization
+import pickle as cPickle
+import tflearn.datasets.mnist as mnist
+import os
+import cv2
+import numpy as np
+import scipy.io.wavfile
+import random
+
+
+
+# length = 20000
+drop_out_prob = 0.5
+def build_tflearn_ann(length):
+    input_layer = input_data(shape=[None, length, 1])
+
+
+    pool_layer_1 = max_pool_1d(input_layer, 10, name='pool_layer_1')
+    pool_layer_2 = max_pool_1d(pool_layer_1, 5, name='pool_layer_2')
+    pool_layer_3 = max_pool_1d(pool_layer_2, 5, name='pool_layer_3')
+    pool_layer_4 = max_pool_1d(pool_layer_3, 5, name='pool_layer_3')
+
+    fully_connect_1 = fully_connected(pool_layer_3, 512, activation='relu', name='fully_connect_1',
+                                      weights_init='xavier', regularizer="L2")
+
+    fully_connect_2 = fully_connected(pool_layer_2, 512, activation='relu', name='fully_connect_2',
+                                      weights_init='xavier', regularizer="L2")
+
+    fully_connect_3 = fully_connected(pool_layer_1, 512, activation='relu', name='fully_connect_3',
+                                      weights_init='xavier', regularizer="L2")
+
+    fully_connect_4 = fully_connected(pool_layer_4, 512, activation='relu', name='fully_connect_3',
+                                      weights_init='xavier', regularizer="L2")
+    # Merge above layers
+    merge_layer = tflearn.merge_outputs([fully_connect_1, fully_connect_2, fully_connect_3, fully_connect_4])
+    # merge_layer = tflearn.merge_outputs(
+    #     [fully_connect_1, fully_connect_2, fully_connect_3, fully_connect_4, fully_connect_5])
+    # merge_layer = tflearn.merge_outputs(
+    #     [fully_connect_1, fully_connect_2, fully_connect_3, fully_connect_4, fully_connect_5, fully_connect_6,
+    #      fully_connect_7, fully_connect_8, fully_connect_9, fully_connect_10])
+    drop_2 = dropout(merge_layer, 0.25)
+
+
+    fc_layer_4 = fully_connected(drop_2, 2048, activation='relu', name='fc_layer_4', regularizer='L2',
+                                 weights_init='xavier', weight_decay=0.001)
+    drop_2 = dropout(fc_layer_4, drop_out_prob)
+
+
+    fc_layer_5 = fully_connected(drop_2, 1024, activation='relu', name='fc_layer_5', regularizer='L2',
+                                 weights_init='xavier', weight_decay=0.001)
+    drop_3 = dropout(fc_layer_5, drop_out_prob)
+
+    fc_layer_6 = fully_connected(drop_3, 128, activation='relu', name='fc_layer_5', regularizer='L2',
+                                 weights_init='xavier', weight_decay=0.001)
+    drop_4 = dropout(fc_layer_6, drop_out_prob)
+
+    # Output
+    fc_layer_2 = fully_connected(drop_4, 3, activation='softmax', name='output')
+    network = regression(fc_layer_2, optimizer='adam', loss='softmax_categorical_crossentropy', learning_rate=0.0001,
+                         metric='Accuracy')
+    model = tflearn.DNN(network)
+    return model

model_audio_CNN.py

@@ -0,0 +1,62 @@
+import numpy as np
+import tensorflow as tf
+import tflearn
+from tflearn.data_utils import shuffle, to_categorical
+from tflearn.layers.core import input_data, dropout, fully_connected, flatten, reshape
+from tflearn.layers.conv import conv_2d, max_pool_2d
+from tflearn.layers.conv import conv_1d, max_pool_1d, avg_pool_1d
+from tflearn.layers.estimator import regression
+from tflearn.layers.normalization import batch_normalization
+import pickle as cPickle
+import tflearn.datasets.mnist as mnist
+import os
+import cv2
+import numpy as np
+import scipy.io.wavfile
+import random
+
+
+
+drop_out_prob = 0.25
+def build_tflearn_cnn(length):
+    input_layer = input_data(shape=[None, length, 1])
+
+
+    # Convolution Layer
+    conv_layer_1 = conv_1d(input_layer,
+                           nb_filter=512,
+                           filter_size=10,
+                           activation='relu',
+                           name='conv_layer_1',
+                           weights_init='xavier',
+                           regularizer="L2")
+    pool_layer_1 = max_pool_1d(conv_layer_1, 4, name='pool_layer_1')
+
+
+    conv_layer_2 = conv_1d(pool_layer_1,
+                           nb_filter=512,
+                           filter_size=5,
+                           activation='relu',
+                           name='conv_layer_2',
+                           weights_init='xavier',
+                           regularizer="L2")
+    pool_layer_3 = max_pool_1d(conv_layer_2, 4, name='pool_layer_3')
+    # flat = flatten(pool_layer_3)
+
+
+    fc_layer_4 = fully_connected(pool_layer_3, 256, activation='relu', name='fc_layer_4', regularizer='L2')
+    drop_2 = dropout(fc_layer_4, drop_out_prob)
+    fc_layer_5 = fully_connected(drop_2, 128, activation='relu', name='fc_layer_5', regularizer='L2')
+    drop_3 = dropout(fc_layer_5, drop_out_prob)
+
+    # Output
+    fc_layer_2 = fully_connected(drop_3, 3, activation='softmax', name='output')
+    network = regression(fc_layer_2, optimizer='adam', loss='softmax_categorical_crossentropy', learning_rate=0.0001,
+                         metric='accuracy')
+    model = tflearn.DNN(network,
+                        tensorboard_verbose=0)
+
+    return model
+
+
+

train_model_ANN_audio.py

@@ -0,0 +1,60 @@
+import pickle as cPickle
+import tensorflow as tf
+import model_audio_ANN
+import numpy as np
+
+
+pickle_path = './pickle_data/'
+print("loading pickle files for ANN")
+with open(pickle_path+"train_data_22k_org.pickle", "rb") as input_file:
+    x_train = cPickle.load(input_file)
+
+with open(pickle_path+"train_labels_22k_org.pickle", "rb") as input_file:
+    y_train = cPickle.load(input_file)
+
+with open(pickle_path+"test_data_22k_org.pickle", "rb") as input_file:
+    x_test = cPickle.load(input_file)
+
+with open(pickle_path+"test_labels_22k_org.pickle", "rb") as input_file:
+    y_test = cPickle.load(input_file)
+
+with open(pickle_path + 'valid_data_22k_org.pickle', 'rb') as input_file:
+    x_valid = cPickle.load(input_file)
+
+with open(pickle_path + "valid_labels_22k_org.pickle", "rb") as input_file:
+    y_valid = cPickle.load(input_file)
+
+
+
+x_train = np.row_stack([x_train, x_test])
+y_train = np.row_stack([y_train, y_test])
+
+
+ann_model_dir = './model/ANN/'
+
+##############
+# Train ANN ##
+##############
+NUM_EPOCHS = 500
+BATCH_SIZE = 64
+MODEL = model_audio_ANN.build_tflearn_ann(x_train.shape[1])
+MODEL.fit(x_train, y_train, n_epoch=NUM_EPOCHS,
+          shuffle=True,
+          validation_set=(x_valid, y_valid),
+          show_metric=True,
+          batch_size=BATCH_SIZE)
+MODEL.save(ann_model_dir+'Bee_audio_ANN.tfl')
+
+print(MODEL.evaluate(x_test, y_test))
+print(MODEL.evaluate(x_train, y_train))
+
+
+
+# tf.reset_default_graph()
+# ann_model_dir = './model/ANN_new/Bee_audio_ANN.tfl'
+# ann_model = model_audio_ANN.build_tflearn_ann(x_test.shape[1])
+# ann_model.load(ann_model_dir, weights_only=True, create_new_session = False)
+# print(ann_model.evaluate(x_test, y_test))
+# print(ann_model.evaluate(x_train, y_train))
+# validation_acc = ann_model.evaluate(x_test, y_test)
+# print(validation_acc)

train_model_CNN_audio.py

@@ -0,0 +1,68 @@
+import model_audio_CNN
+import pickle as cPickle
+import tensorflow as tf
+import numpy as np
+import os
+
+os.environ["CUDA_VISIBLE_DEVICES"]="0,1"
+# export CUDA_VISIBLE_DEVICE=0,1
+
+pickle_path = './pickle_data/'
+print("loading pickle files for CNN")
+with open(pickle_path+"train_data_22k_org.pickle", "rb") as input_file:
+    x_train = cPickle.load(input_file)
+
+with open(pickle_path+"train_labels_22k_org.pickle", "rb") as input_file:
+    y_train = cPickle.load(input_file)
+
+with open(pickle_path+"test_data_22k_org.pickle", "rb") as input_file:
+    x_test = cPickle.load(input_file)
+
+with open(pickle_path+"test_labels_22k_org.pickle", "rb") as input_file:
+    y_test = cPickle.load(input_file)
+
+
+x_train = np.row_stack([x_train, x_test])
+y_train = np.row_stack([y_train, y_test])
+
+
+with open(pickle_path + 'valid_data_22k_org.pickle', 'rb') as input_file:
+    x_valid = cPickle.load(input_file)
+
+# with open(pickle_path+"test_labels_22k_org.pickle", "rb") as input_file:
+with open(pickle_path + "valid_labels_22k_org.pickle", "rb") as input_file:
+    y_valid = cPickle.load(input_file)
+
+
+cnn_model_dir = './model/CNN/'
+
+##############
+# Train CNN ##
+##############
+NUM_EPOCHS = 500
+BATCH_SIZE = 8
+MODEL = model_audio_CNN.build_tflearn_cnn(x_train.shape[1])
+# with tf.device('/gpu:0'):
+MODEL.fit(x_train, y_train, n_epoch=NUM_EPOCHS,
+              shuffle=True,
+              validation_set=(x_valid, y_valid),
+              show_metric=True,
+              batch_size=BATCH_SIZE,
+          run_id = 'Bee_audio_CNN_best_4')
+MODEL.save(cnn_model_dir+'Bee_audio_CNN.tfl')
+
+
+
+
+
+# ##################
+# ## Evaluate CNN ##
+# ##################
+# tf.reset_default_graph()
+
+# cnn_model_dir = '/home/vishal/PycharmProjects/bee_audio_project/model/CNN/Bee_audio_CNN_100.tfl'
+# # cnn_model = model_audio_CNN.build_tflearn_cnn(x_test.shape[1])
+
+# MODEL.load(cnn_model_dir, weights_only=True)
+# validation_acc = MODEL.evaluate(x_test, y_test)
+# print(validation_acc)