Merge branch 'master' of gitlab.insa-rouen.fr:mdavid/prj_deep

mobilnet_test.ipynb

+%% Cell type:code id:personal-marshall tags:
+
+``` python
+import numpy as np
+import os
+import pandas as pd
+from scipy.io import wavfile
+
+import librosa
+from tqdm import tqdm
+import tensorflow as tf
+from tensorflow import keras
+
+import matplotlib.pyplot as plt
+from tensorflow.keras import regularizers, activations
+from tensorflow.keras.models import Sequential
+from tensorflow.keras.layers import Dense, Dropout, Flatten, Activation, Conv2D, MaxPooling2D, GlobalAveragePooling2D
+from tensorflow.keras.utils import to_categorical
+
+from tensorflow.keras.preprocessing.image import ImageDataGenerator
+
+from datetime import datetime
+
+from matplotlib import pyplot as plt
+
+from sklearn.model_selection import train_test_split
+import cv2
+```
+
+%% Cell type:code id:intimate-property tags:
+
+``` python
+us8k_df = pd.read_pickle("us8k_df.pkl")
+```
+
+%% Cell type:code id:closed-abraham tags:
+
+``` python
+df = us8k_df.drop(['fold'],axis=1)
+X = np.stack(df.melspectrogram.to_numpy())
+X_dim = (128,128,1)
+X = X.reshape(X.shape[0], *X_dim)
+Y = np.array(df['label'])
+Y = to_categorical(Y)
+```
+
+%% Cell type:code id:controlling-overhead tags:
+
+``` python
+Y.shape, X.shape
+```
+
+%%%% Output: execute_result
+
+    ((8732, 10), (8732, 128, 128, 1))
+
+%% Cell type:code id:dying-cooking tags:
+
+``` python
+X_new = np.zeros((8732,128,128,3))
+```
+
+%% Cell type:code id:increasing-annotation tags:
+
+``` python
+X_new.shape
+```
+
+%%%% Output: execute_result
+
+    (8732, 128, 128, 3)
+
+%% Cell type:code id:rubber-hanging tags:
+
+``` python
+for i in range(len(X)):
+    X_new[i]=cv2.cvtColor(X[i], cv2.COLOR_GRAY2RGB)
+```
+
+%% Cell type:code id:after-driving tags:
+
+``` python
+X=X_new
+```
+
+%% Cell type:code id:illegal-partner tags:
+
+``` python
+X_train, X_test, Y_train, Y_test = train_test_split(X,Y,test_size=0.3,shuffle=True,stratify = Y)
+X_val, X_test, Y_val,Y_test = train_test_split(X_test,Y_test,test_size=0.5,shuffle=True,stratify = Y_test)
+```
+
+%% Cell type:markdown id:french-assembly tags:
+
+MOdèle utilisant mobile net
+
+%% Cell type:code id:geographic-diving tags:
+
+``` python
+preprocess_input = tf.keras.applications.mobilenet_v2.preprocess_input
+rescale = tf.keras.layers.experimental.preprocessing.Rescaling(1./127.5, offset= -1)
+```
+
+%% Cell type:code id:pacific-registration tags:
+
+``` python
+IMG_SHAPE = (128,128,3)
+
+base_model = tf.keras.applications.MobileNetV2(input_shape=IMG_SHAPE,
+                                               include_top=False,
+                                               weights='imagenet')
+```
+
+%% Cell type:code id:included-mozambique tags:
+
+``` python
+prediction_layer = tf.keras.layers.Dense(10)
+global_average_layer = tf.keras.layers.GlobalAveragePooling2D()
+```
+
+%% Cell type:code id:palestinian-crowd tags:
+
+``` python
+# model : MobileNet puis du dropout et une couche dense pour la prédiction
+inputs = tf.keras.Input(shape=IMG_SHAPE)
+x = preprocess_input(inputs)
+x = rescale(x)
+x = base_model(x, training=False)
+x = global_average_layer(x)
+x = tf.keras.layers.Dropout(0.2)(x)
+outputs = prediction_layer(x)
+model = tf.keras.Model(inputs, outputs)
+```
+
+%% Cell type:code id:simple-rebecca tags:
+
+``` python
+model.summary()
+```
+
+%%%% Output: stream
+
+    Model: "model"
+    _________________________________________________________________
+    Layer (type)                 Output Shape              Param #
+    =================================================================
+    input_2 (InputLayer)         [(None, 128, 128, 3)]     0
+    _________________________________________________________________
+    tf.math.truediv (TFOpLambda) (None, 128, 128, 3)       0
+    _________________________________________________________________
+    tf.math.subtract (TFOpLambda (None, 128, 128, 3)       0
+    _________________________________________________________________
+    rescaling (Rescaling)        (None, 128, 128, 3)       0
+    _________________________________________________________________
+    mobilenetv2_1.00_128 (Functi (None, 4, 4, 1280)        2257984
+    _________________________________________________________________
+    global_average_pooling2d (Gl (None, 1280)              0
+    _________________________________________________________________
+    dropout (Dropout)            (None, 1280)              0
+    _________________________________________________________________
+    dense (Dense)                (None, 10)                12810
+    =================================================================
+    Total params: 2,270,794
+    Trainable params: 2,236,682
+    Non-trainable params: 34,112
+    _________________________________________________________________
+
+%% Cell type:code id:pacific-correction tags:
+
+``` python
+base_learning_rate = 0.001
+model.compile(optimizer=tf.keras.optimizers.Adam(lr=base_learning_rate),
+              loss=tf.keras.losses.BinaryCrossentropy(from_logits=True),
+              metrics=['accuracy'])
+```
+
+%% Cell type:code id:durable-straight tags:
+
+``` python
+initial_epochs = 10
+num_batch_size = 32
+loss0, accuracy0 = model.evaluate(X_val,Y_val)
+```
+
+%%%% Output: stream
+
+    41/41 [==============================] - 5s 99ms/step - loss: 0.9276 - accuracy: 0.1113
+
+%% Cell type:code id:victorian-hawaiian tags:
+
+``` python
+log_dir = "logs/fit/" + datetime.now().strftime("%Y%m%d-%H%M%S")
+tensorboard_callback = tf.keras.callbacks.TensorBoard(log_dir = log_dir, histogram_freq = 1)
+save_best = tf.keras.callbacks.ModelCheckpoint(filepath = "logs/checkpoints/", save_weights_only = True,
+                                               monitor = "val_accuracy", mode = "max", save_best_only = True)
+```
+
+%% Cell type:code id:adolescent-prerequisite tags:
+
+``` python
+model_fit = model.fit(X_train[:2000],Y_train[0:2000], epochs=initial_epochs,validation_data=(X_val[0:300],Y_val[0:300]),batch_size=num_batch_size,callbacks = [tensorboard_callback, save_best])
+```
+
+%%%% Output: stream
+
+    Epoch 1/10
+    63/63 [==============================] - 37s 552ms/step - loss: 0.4256 - accuracy: 0.1145 - val_loss: 0.3422 - val_accuracy: 0.1133
+    Epoch 2/10
+     4/63 [>.............................] - ETA: 39s - loss: 0.3885 - accuracy: 0.0859
+
+%% Cell type:code id:civic-trinity tags:
+
+``` python
+```