Updated file

Image_Captioning.ipynb

@@ -599,9 +599,7 @@
         "id": "lF4e-SWgmH-R"
       },
       "source": [
-        "You can download the features file from [here](#). Make sure to move it into the `export_model` folder. The file must be labelled as `features.pkl`.\n",
-        "\n",
-        "**CHANGE THIS**"
+        "You can download the features file from [here](https://github.com/yanukadeneth99/Image-Captioning-DL/blob/master/Resources/features.pkl). Make sure to move it into the `export_model` folder. The file must be labelled as `features.pkl`."
       ]
     },
     {
@@ -1156,7 +1154,9 @@
       "source": [
         "### Loading Weights\n",
         "\n",
-        "We can load the weights if they are saved as well. Make sure the file name is `main_model_weights.hdf5` and is under the `export_model` folder."
+        "We can load the weights if they are saved as well. Make sure the file name is `main_model_weights.hdf5` and is under the `export_model` folder.\n",
+        "\n",
+        "You can download the weights from [here](https://github.com/yanukadeneth99/Image-Captioning-DL/blob/master/Resources/main_model_weights.hdf5)."
       ]
     },
     {
@@ -1834,21 +1834,96 @@
     },
     {
       "cell_type": "markdown",
+      "source": [
+        "We can see that the guesses the model makes are pretty close and will improve once we train more. \n",
+        "\n",
+        "However, moving on, let's setup a way so that you can test with your own image as well.\n",
+        "\n",
+        "# Testing\n",
+        "\n",
+        "We can upload our custom image to the `test` folder and set the image name in the below variable and run to test."
+      ],
       "metadata": {
-        "id": "YEInN-Pb3r25"
+        "id": "MYpL53JEGEUO"
+      }
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "# Create the test folder\n",
+        "!mkdir test"
+      ],
+      "metadata": {
+        "id": "sxtEUZ9iILuT"
       },
-      "source": []
+      "execution_count": null,
+      "outputs": []
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "# Set Image to Test\n",
+        "test_image_path = \"\""
+      ],
+      "metadata": {
+        "id": "k05niV1TGQLq"
+      },
+      "execution_count": null,
+      "outputs": []
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "# Load Image\n",
+        "t_image = load_img(test_image_path, target_size=(224, 224))\n",
+        "# Covert Image to Numpy Array\n",
+        "t_image = img_to_array(t_image)\n",
+        "# Reshape for Model\n",
+        "t_image = t_image.reshape((1, image.shape[0], image.shape[1], image.shape[2]))\n",
+        "# Preprocess the values\n",
+        "t_image = preprocess_input(t_image)\n",
+        "# Extract features using the model\n",
+        "t_feature = vgg_model.predict(t_image, verbose=0)\n",
+        "\n",
+        "# Predict the caption from the extracted features\n",
+        "t_caption = predict_caption(model, t_feature, tokenizer, max_length)\n",
+        "\n",
+        "# Print the caption\n",
+        "plt.imshow(image)\n",
+        "print(f\"Caption Generated : {t_caption}\")"
+      ],
+      "metadata": {
+        "id": "7wiAl6GDGayu"
+      },
+      "execution_count": null,
+      "outputs": []
+    },
+    {
+      "cell_type": "markdown",
+      "source": [
+        "# Conclusion\n",
+        "\n",
+        "We have created a Neural Network for Image Captioning from Scratch. We can train the model more to perform better and tweak around few values to find the best possible model for this."
+      ],
+      "metadata": {
+        "id": "-31QNwVaHMN5"
+      }
     },
     {
       "cell_type": "markdown",
       "metadata": {
         "id": "D04bx2U9nVCw"
       },
       "source": [
-        "# Credits\n",
+        "# Links\n",
+        "\n",
+        "- [GitHub Project](https://github.com/yanukadeneth99/Image-Captioning-DL)\n",
+        "- [Twitter](https://twitter.com/yanukadeneth99)\n",
+        "\n",
+        "# Resources\n",
         "\n",
         "- [VGG16 → from scratch using Transfer Learning with Keras and TensorFlow 2](https://narenltk.medium.com/vgg16-from-scratch-using-transfer-learning-with-keras-and-tensorflow-ffee5ae460fb)\n",
-        "- "
+        "- [Image Caption Generator using Python | Flickr Dataset | Deep Learning Tutorial](https://www.hackersrealm.net/post/image-caption-generator-using-python)"
       ]
     }
   ],