Jupyter notebook to run inference on keras's classification model on a folder of images, and display the classification results with images for interpretation.

keras_model_inference.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "source": [
    "import tensorflow as tf\n",
    "import numpy as np \n",
    "from tensorflow.keras.preprocessing import image\n",
    "import os\n",
    "import matplotlib.pyplot as plt"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "markdown",
   "source": [
    "## Configurations"
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "source": [
    "DATA_FOLDER = \"data\"\n",
    "# folder containing the classes\n",
    "data_dir = \"./data/{}\".format(DATA_FOLDER)\n",
    "MODEL_NAME=\"model_name\"\n",
    "model_dir=\"./exported_models/{}\".format(MODEL_NAME)\n",
    "\n",
    "class_indices={0:\"class1\", 1:\"class2\",2:\"class3\"}\n",
    "inv_classes = {v: k for k, v in class_indices.items()}\n",
    "num_classes=len(class_indices)\n",
    "target_size=(128,128)"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "markdown",
   "source": [
    "## Function definitions"
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "source": [
    "def plot_image(i, predictions_array, true_label, img):\n",
    "  true_label, img = true_label[i], img[i].astype('uint8')\n",
    "  plt.grid(False)\n",
    "  plt.xticks([])\n",
    "  plt.yticks([])\n",
    "\n",
    "  plt.imshow(img, cmap=plt.cm.binary)\n",
    "\n",
    "  predicted_label = np.argmax(predictions_array)\n",
    "  if predicted_label == true_label:\n",
    "    color = 'blue'\n",
    "  else:\n",
    "    color = 'red'\n",
    "\n",
    "  plt.xlabel(\"{} {:2.0f}% ({})\".format(class_indices[predicted_label],\n",
    "                                100*np.max(predictions_array),\n",
    "                                class_indices[true_label]),\n",
    "                                color=color)\n",
    "\n",
    "def plot_value_array(i, predictions_array, true_label):\n",
    "  true_label = true_label[i]\n",
    "  plt.grid(False)\n",
    "  plt.xticks(range(num_classes))\n",
    "  plt.yticks([])\n",
    "  thisplot = plt.bar(range(num_classes), predictions_array, color=\"#777777\")\n",
    "  plt.ylim([0, 1])\n",
    "  predicted_label = np.argmax(predictions_array)\n",
    "\n",
    "  thisplot[predicted_label].set_color('red')\n",
    "  thisplot[true_label].set_color('blue')\n"
   ],
   "outputs": [],
   "metadata": {}
  },
  {
   "cell_type": "markdown",
   "source": [
    "## Run"
   ],
   "metadata": {}
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "source": [
    "# load model\n",
    "model = tf.keras.models.load_model(model_dir)\n",
    "\n",
    "#loop for all classes\n",
    "for (root,dirs,files) in os.walk(data_dir):\n",
    "  if len(files) >0:\n",
    "    # load images\n",
    "    img_list=[]\n",
    "    test_labels=[]\n",
    "    test_images=[]\n",
    "    for f in files:\n",
    "      if f.endswith(\".jpg\") or f.endswith('.png'):\n",
    "        img = image.load_img(os.path.join(root,f), target_size=target_size)\n",
    "        x = image.img_to_array(img)\n",
    "        test_images.append(x)\n",
    "        x = np.expand_dims(x, axis=0)\n",
    "        img_list.append(x)\n",
    "        test_labels.append(inv_classes[os.path.basename(root)])\n",
    "    images=np.vstack(img_list)\n",
    "    # predict\n",
    "    predictions = model.predict(images)\n",
    "\n",
    "    # show predictions\n",
    "    num_rows = 5\n",
    "    num_cols = 5\n",
    "    num_images = num_rows*num_cols\n",
    "    plt.figure(figsize=(2*2*num_cols, 2*num_rows))\n",
    "    for i in range(num_images):\n",
    "      plt.subplot(num_rows, 2*num_cols, 2*i+1)\n",
    "      plot_image(i, predictions[i], test_labels, test_images)\n",
    "      plt.subplot(num_rows, 2*num_cols, 2*i+2)\n",
    "      plot_value_array(i, predictions[i], test_labels)\n",
    "    plt.tight_layout()\n",
    "    plt.show()"
   ],
   "outputs": [],
   "metadata": {}
  }
 ],
 "metadata": {
  "orig_nbformat": 4,
  "language_info": {
   "name": "python",
   "version": "3.6.9",
   "mimetype": "text/x-python",
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "pygments_lexer": "ipython3",
   "nbconvert_exporter": "python",
   "file_extension": ".py"
  },
  "kernelspec": {
   "name": "python3",
   "display_name": "Python 3.6.9 64-bit ('keras': venv)"
  },
  "interpreter": {
   "hash": "caa3c448db030c47d443f1bf3e4d2b22169e43afde0cc3d48598503fa91d349c"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}