imageclassification

romdhanisam · Jan 21, 2018 · d96ff54 · d96ff54
1 parent ef7e251
commit d96ff54
Show file tree

Hide file tree

Showing 5 changed files with 97 additions and 481 deletions.
diff --git a/.ipynb_checkpoints/ImageClassification-checkpoint.ipynb b/.ipynb_checkpoints/ImageClassification-checkpoint.ipynb
@@ -120,7 +120,9 @@
   {
    "cell_type": "code",
    "execution_count": 156,
-   "metadata": {},
+   "metadata": {
+    "collapsed": true
+   },
    "outputs": [],
    "source": [
     "# Importation Conv2D de keras.layers, \n",
@@ -172,7 +174,9 @@
   {
    "cell_type": "code",
    "execution_count": 159,
-   "metadata": {},
+   "metadata": {
+    "collapsed": true
+   },
    "outputs": [],
    "source": [
     "# Importation Flatten de keras.layers, qui est utilisé pour \n",
@@ -637,48 +641,14 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 15,
+   "execution_count": null,
    "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "array([[0.4838731 ],\n",
-       "       [0.5006337 ],\n",
-       "       [0.46920896],\n",
-       "       ...,\n",
-       "       [0.486258  ],\n",
-       "       [0.44625363],\n",
-       "       [0.47998008]], dtype=float32)"
-      ]
-     },
-     "execution_count": 15,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
+   "outputs": [],
    "source": [
     "#fit_generator(self, generator, steps_per_epoch=None,epochs=1, verbose=1, callbacks=None,validation_data=None, validation_steps=None, class_weight=None, max_queue_size=10, workers=1, use_multiprocessing=False, shuffle=True, initial_epoch=0)\n",
-    "\n",
     "a = model.predict_generator(train_generator, steps=None, max_queue_size=10, workers=1, use_multiprocessing=False, verbose=0)\n",
     "a"
    ]
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "# Réseau pré-entrainé : VGG16"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {
-    "collapsed": true
-   },
-   "outputs": [],
-   "source": []
   }
  ],
  "metadata": {

diff --git a/Feuille de calcul sans titre - Feuille 1.pdf → Feuille de calcul.pdf b/Feuille de calcul sans titre - Feuille 1.pdf → Feuille de calcul.pdf
diff --git a/ImageClassification.ipynb b/ImageClassification.ipynb
@@ -120,7 +120,9 @@
   {
    "cell_type": "code",
    "execution_count": 156,
-   "metadata": {},
+   "metadata": {
+    "collapsed": true
+   },
    "outputs": [],
    "source": [
     "# Importation Conv2D de keras.layers, \n",
@@ -172,7 +174,9 @@
   {
    "cell_type": "code",
    "execution_count": 159,
-   "metadata": {},
+   "metadata": {
+    "collapsed": true
+   },
    "outputs": [],
    "source": [
     "# Importation Flatten de keras.layers, qui est utilisé pour \n",
@@ -637,48 +641,14 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 15,
+   "execution_count": null,
    "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "array([[0.4838731 ],\n",
-       "       [0.5006337 ],\n",
-       "       [0.46920896],\n",
-       "       ...,\n",
-       "       [0.486258  ],\n",
-       "       [0.44625363],\n",
-       "       [0.47998008]], dtype=float32)"
-      ]
-     },
-     "execution_count": 15,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
+   "outputs": [],
    "source": [
     "#fit_generator(self, generator, steps_per_epoch=None,epochs=1, verbose=1, callbacks=None,validation_data=None, validation_steps=None, class_weight=None, max_queue_size=10, workers=1, use_multiprocessing=False, shuffle=True, initial_epoch=0)\n",
-    "\n",
     "a = model.predict_generator(train_generator, steps=None, max_queue_size=10, workers=1, use_multiprocessing=False, verbose=0)\n",
     "a"
    ]
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "# Réseau pré-entrainé : VGG16"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {
-    "collapsed": true
-   },
-   "outputs": [],
-   "source": []
   }
  ],
  "metadata": {

diff --git a/ImageClassification.py b/ImageClassification.py
@@ -0,0 +1,81 @@
+from keras.models import Sequential
+from keras.layers import Conv2D
+from keras.layers import MaxPooling2D
+from keras.layers import Flatten
+from keras.layers import Dense
+from keras.layers import Activation
+from keras.layers import Dropout
+import keras.optimizers as ko
+
+import tensorflow as tf
+import keras.backend as k
+
+data_dir = 'data'
+N_train = 1000
+N_val = 400
+
+img_width = 150
+img_height = 150
+
+model = Sequential()
+# Importation Conv2D de keras.layers, 
+# ceci pour effectuer l'opération de convolution, 
+model.add(Conv2D(32, (3, 3), input_shape=(img_width, img_height, 3), data_format="channels_last"))
+model.add(Activation('relu'))
+#importation MaxPooling2D de keras.layers, qui est utilisé pour l'opération de pooling
+# Nous utilisons une fonction de Maxpool
+model.add(MaxPooling2D(pool_size=(2, 2)))
+
+# Importation Flatten de keras.layers, qui est utilisé pour 
+#  la conversion de tous les tableaux bidimensionnels résultants en un seul vecteur.
+model.add(Flatten())
+# Importation Dense de keras.layers, qui est utilisé pour effectuer "the full connection of the neural network"
+model.add(Dense(64,name='first', input_shape=(img_width, img_height, 3)))
+model.add(Activation('relu'))
+model.add(Dropout(0.5))
+model.add(Dense(1,name='second'))
+model.add(Activation('sigmoid'))
+
+model.compile(loss = 'binary_crossentropy',
+              optimizer = 'adam',
+              metrics = ['accuracy'])
+
+import keras.preprocessing.image as kpi
+train_datagen = kpi.ImageDataGenerator(
+        rescale=1./255,
+        shear_range=0.2,
+        zoom_range=0.2,
+        horizontal_flip=True)
+valid_datagen = kpi.ImageDataGenerator(rescale=1./255)
+
+#Nous divisons l'ensemble d'entraînement en batchs, 
+#chaque epochs passe par tout l'ensemble d'entraînement.
+#Chaque itération passe par batch.
+batch_size = 100 # 1000/100
+epochs = 50
+
+#Un générateur qui va lire les images trouvées dans 'data / train'
+train_generator = train_datagen.flow_from_directory(
+        data_dir+"/train/",  #le répertoire
+        target_size=(img_width, img_height), #toutes les images seront redimensionnées à 150x150
+        batch_size=batch_size,
+        class_mode='binary',
+        classes=['cats','dogs'])
+# Un générateur similaire, pour les données de validation
+validation_generator = valid_datagen.flow_from_directory(
+        data_dir+"/validation/",
+        target_size=(img_width, img_height),
+        batch_size=batch_size,
+        class_mode='binary')
+model.fit_generator(train_generator,
+                    steps_per_epoch=N_train// batch_size,
+                    epochs=epochs,
+                    validation_data=validation_generator,
+                    validation_steps=N_val// batch_size)
+#
+model.save('data\\model\\models_convolutional_network_%d_epochs_%d_batch_size.h5' %(epochs, batch_size))
+#
+score_conv_train = model.evaluate_generator(train_generator, N_train// batch_size)
+score_conv_val = model.evaluate_generator(validation_generator, N_val //batch_size)
+print('Train accuracy:', score_conv_train[1])
+print('Test accuracy:', score_conv_val[1])