side view model added

Backend/SF_Weed_identification_Finalized.ipynb

@@ -26,7 +26,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "dataset_location  = '/content/drive/MyDrive/RP_SmartFarmer/Sandhini Gamage - Weed identification /Dataset/FinalizedWeedDataSet.zip'"
+    "dataset_location  = '/content/drive/MyDrive/FinalizedWeedDataSet.zip'"
    ]
   },
   {
@@ -55,6 +55,216 @@
     "for dirpath , dirnames , filenames in os.walk(FILE_DIR):\n",
     "  print(f\"There are {len(dirnames)} directories and {len(filenames)} images in '{dirpath}'.\")"
    ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "✅ 01- A Model for Side View"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "train_dir_side = \"/content/content/Weed Dataset/Side view/training\" #Training Directory path of the Side View Image Dataset\n",
+    "test_dir_side = \"/content/content/Weed Dataset/Side view/testing\"\n",
+    "\n",
+    "#lets get the class names\n",
+    "import pathlib\n",
+    "import numpy as np\n",
+    "\n",
+    "data_dir = pathlib.Path(train_dir_side)\n",
+    "class_names = np.array(sorted([item.name for item in data_dir.glob('*')]))\n",
+    "print(class_names)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "🟨 Data Visualization"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import matplotlib.pyplot as plt\n",
+    "import matplotlib.image as mpimg\n",
+    "import random\n",
+    "import os\n",
+    "\n",
+    "def view_random_image(target_dir , target_class):\n",
+    "  ''' \n",
+    "    This Function will displays a random Image from the dataset\n",
+    "  '''\n",
+    "  target_folder = target_dir+\"/\"+target_class\n",
+    "  random_image = random.sample(os.listdir(target_folder) ,1)\n",
+    "\n",
+    "  img = mpimg.imread(target_folder+\"/\"+random_image[0])\n",
+    "  plt.imshow(img)\n",
+    "  plt.title(target_class)\n",
+    "  plt.axis(\"off\")\n",
+    "\n",
+    "  print(f\"Image Shape:{img.shape}\") \n",
+    "\n",
+    "  return img"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import random\n",
+    "img = view_random_image(target_dir=train_dir_side , \n",
+    "                        target_class = random.choice(class_names))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "🟨 DATA Preprocessing"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
+    "\n",
+    "#Creating Image Data Generators for Training Data with augmentation\n",
+    "train_data_gen  = ImageDataGenerator(rescale=1/255.,\n",
+    "                                     #preprocessing_function=to_grayscale_then_rgb,\n",
+    "                                     rotation_range = 0.2,\n",
+    "                                     shear_range  = 0.2,\n",
+    "                                     zoom_range = 0.2,\n",
+    "                                     width_shift_range=0.3,\n",
+    "                                     height_shift_range = 0.3,\n",
+    "                                     horizontal_flip= True)\n",
+    "\n",
+    "#Create ImageDatagenerator for testing data\n",
+    "test_data_gen = ImageDataGenerator(rescale=1/255.)\n",
+    "\n",
+    "#Import and Transform/pre process the data\n",
+    "train_data_multi = train_data_gen.flow_from_directory(train_dir_side,\n",
+    "                                                      target_size = (224,224),\n",
+    "                                                      batch_size = 32,\n",
+    "                                                      class_mode = 'categorical',\n",
+    "                                                      shuffle = True)\n",
+    "test_data_multi = test_data_gen.flow_from_directory(test_dir_side,\n",
+    "                                                    target_size = (224, 224),\n",
+    "                                                    batch_size =  32,\n",
+    "                                                    class_mode = 'categorical', #This will gives us one-hot encoded data\n",
+    "                                                    shuffle = True)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "🟨 EfficientNetB0 Model"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "efficientnet_url = \"https://tfhub.dev/tensorflow/efficientnet/b0/feature-vector/1\""
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "#Import Dependencies\n",
+    "import tensorflow as tf\n",
+    "import tensorflow_hub as hub\n",
+    "from tensorflow.keras import layers"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "#Lets make a create_model() function to create a model from a URL\n",
+    "def create_model(model_url , num_classes=7):\n",
+    "\n",
+    "  feature_extractor_layer = hub.KerasLayer(model_url,\n",
+    "                                           trainable = False, #freeze the already learned patterns \n",
+    "                                           name = \"feature_extraction_layer\",\n",
+    "                                           input_shape = (224, 224,3)) \n",
+    "  #Create our own model\n",
+    "  model  =  tf.keras.Sequential([\n",
+    "    feature_extractor_layer,\n",
+    "    layers.Dense(num_classes , activation=\"softmax\" , name=\"output_layer\")\n",
+    "  ])\n",
+    "\n",
+    "  return model"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "#Create Resnet Model\n",
+    "efficientnet_model_side =  create_model(efficientnet_url , \n",
+    "                             num_classes = 3)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "#Compile our resnet model\n",
+    "efficientnet_model_side.compile(loss='categorical_crossentropy',\n",
+    "                     optimizer = tf.keras.optimizers.Adam(),\n",
+    "                     metrics=[\"accuracy\"])\n",
+    "#Fitting the model\n",
+    "side_vw_model_hist = efficientnet_model_side.fit(train_data_multi,\n",
+    "                               epochs=8,\n",
+    "                               steps_per_epoch=len(train_data_multi),\n",
+    "                               validation_data = test_data_multi,\n",
+    "                               validation_steps = len(test_data_multi)\n",
+    "                               )"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "#Saving the h5 model back to the google drive\n",
+    "efficientnet_model_side.save('/content/drive/MyDrive/RP_SmartFarmer/Sandhini Gamage - Weed identification /private/saved_models/Weed_side_view2.h5')"
+   ]
   }
  ],
  "metadata": {