From a18cb82f341c330a7865dc49bf141e2660d1f1f3 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Aur=C3=A9lien=20Geron?= <ageron@users.noreply.github.com>
Date: Mon, 31 Jan 2022 11:10:02 +1300
Subject: [PATCH] Remove transpose conv section, fix python 3.7->3.8, small
 tweaks

---
 14_deep_computer_vision_with_cnns.ipynb | 48 ++++++++++++-------------
 1 file changed, 22 insertions(+), 26 deletions(-)

diff --git a/14_deep_computer_vision_with_cnns.ipynb b/14_deep_computer_vision_with_cnns.ipynb
index 6d4ba18..ad810d4 100644
--- a/14_deep_computer_vision_with_cnns.ipynb
+++ b/14_deep_computer_vision_with_cnns.ipynb
@@ -63,7 +63,7 @@
    "source": [
     "import sys\n",
     "\n",
-    "assert sys.version_info >= (3, 7)"
+    "assert sys.version_info >= (3, 8)"
    ]
   },
   {
@@ -557,7 +557,7 @@
    },
    "outputs": [],
    "source": [
-    "# extra code – this cells generates and saves Figure 14–9\n",
+    "# extra code – this cells shows what max pooling with stride = 2 looks like\n",
     "\n",
     "import matplotlib as mpl\n",
     "\n",
@@ -572,7 +572,6 @@
     "ax2.set_title(\"Output\", fontsize=14)\n",
     "ax2.imshow(output[0])  # plot the output for the 1st image\n",
     "ax2.axis(\"off\")\n",
-    "save_fig(\"china_max_pooling\")\n",
     "plt.show()"
    ]
   },
@@ -1436,21 +1435,6 @@
   {
    "cell_type": "code",
    "execution_count": 57,
-   "metadata": {
-    "id": "E0XZoWKjpKz_"
-   },
-   "outputs": [],
-   "source": [
-    "def add_random_bounding_boxes(images, labels):\n",
-    "    fake_bboxes = tf.random.uniform([tf.shape(images)[0], 4])\n",
-    "    return images, (labels, fake_bboxes)\n",
-    "\n",
-    "fake_train_set = train_set.take(5).repeat(2).map(add_random_bounding_boxes)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 58,
    "metadata": {
     "colab": {
      "base_uri": "https://localhost:8080/"
@@ -1460,6 +1444,14 @@
    },
    "outputs": [],
    "source": [
+    "# extra code – fits the model using random target bounding boxes (in real life\n",
+    "#              you would need to create proper targets instead)\n",
+    "\n",
+    "def add_random_bounding_boxes(images, labels):\n",
+    "    fake_bboxes = tf.random.uniform([tf.shape(images)[0], 4])\n",
+    "    return images, (labels, fake_bboxes)\n",
+    "\n",
+    "fake_train_set = train_set.take(5).repeat(2).map(add_random_bounding_boxes)\n",
     "model.fit(fake_train_set, epochs=2)"
    ]
   },
@@ -1474,7 +1466,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 59,
+   "execution_count": 58,
    "metadata": {
     "id": "fgjxsrkLpKz_"
    },
@@ -1486,7 +1478,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 60,
+   "execution_count": 59,
    "metadata": {
     "colab": {
      "base_uri": "https://localhost:8080/",
@@ -1579,7 +1571,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 61,
+   "execution_count": 60,
    "metadata": {
     "colab": {
      "base_uri": "https://localhost:8080/"
@@ -1589,7 +1581,8 @@
    },
    "outputs": [],
    "source": [
-    "(X_train_full, y_train_full), (X_test, y_test) = tf.keras.datasets.mnist.load_data()\n",
+    "mnist = tf.keras.datasets.mnist.load_data()\n",
+    "(X_train_full, y_train_full), (X_test, y_test) = mnist\n",
     "X_train_full = X_train_full / 255.\n",
     "X_test = X_test / 255.\n",
     "X_train, X_valid = X_train_full[:-5000], X_train_full[-5000:]\n",
@@ -1602,7 +1595,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 62,
+   "execution_count": 61,
    "metadata": {
     "colab": {
      "base_uri": "https://localhost:8080/"
@@ -1617,12 +1610,15 @@
     "np.random.seed(42)\n",
     "\n",
     "model = tf.keras.Sequential([\n",
-    "    tf.keras.layers.Conv2D(32, kernel_size=3, padding=\"same\", activation=\"relu\"),\n",
-    "    tf.keras.layers.Conv2D(64, kernel_size=3, padding=\"same\", activation=\"relu\"),\n",
+    "    tf.keras.layers.Conv2D(32, kernel_size=3, padding=\"same\",\n",
+    "                           activation=\"relu\", kernel_initializer=\"he_normal\"),\n",
+    "    tf.keras.layers.Conv2D(64, kernel_size=3, padding=\"same\",\n",
+    "                           activation=\"relu\", kernel_initializer=\"he_normal\"),\n",
     "    tf.keras.layers.MaxPool2D(),\n",
     "    tf.keras.layers.Flatten(),\n",
     "    tf.keras.layers.Dropout(0.25),\n",
-    "    tf.keras.layers.Dense(128, activation=\"relu\"),\n",
+    "    tf.keras.layers.Dense(128, activation=\"relu\",\n",
+    "                          kernel_initializer=\"he_normal\"),\n",
     "    tf.keras.layers.Dropout(0.5),\n",
     "    tf.keras.layers.Dense(10, activation=\"softmax\")\n",
     "])\n",