vision-studio-demo.ipynb

{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Vision Studio Demo\n",
    "\n",
    "### Pre-requisites\n",
    "- Δημιουργία ενός Computer Vision resource στο Azure\n",
    "- Δημιουργία ενός .env αρχείου με τα στοιχεία του resource (Endpoint, Key) στον φάκελο του project"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Authentication\n",
    "%pip install azure-ai-vision pillow matplotlib\n",
    "\n",
    "import azure.ai.vision as sdk\n",
    "import os\n",
    "from dotenv import load_dotenv\n",
    "from PIL import Image\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.patches as patches\n",
    "\n",
    "load_dotenv()\n",
    "service_options = sdk.VisionServiceOptions(os.getenv('CV_ENDPOINT'), os.getenv('CV_KEY'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Επιλογή εικόνας για ανάλυση\n",
    "original_image = 'images/turtle-coral-reef.jpg'\n",
    "vision_source = sdk.VisionSource(filename=original_image)\n",
    "\n",
    "# Επιλογή τύπου ανάλυσης\n",
    "\n",
    "## Αφαίρεση του background\n",
    "\n",
    "### Δημιουργία caption για την εικόνα\n",
    "analysis_options = sdk.ImageAnalysisOptions()\n",
    "\n",
    "analysis_options.segmentation_mode = sdk.ImageSegmentationMode.BACKGROUND_REMOVAL\n",
    "\n",
    "# Εμφάνιση της αρχικής εικόνας\n",
    "plt.imshow(Image.open(original_image))\n",
    "plt.axis(\"off\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "image_analyzer = sdk.ImageAnalyzer(service_options, vision_source, analysis_options)\n",
    "\n",
    "result = image_analyzer.analyze()\n",
    "\n",
    "if result.reason == sdk.ImageAnalysisResultReason.ANALYZED:\n",
    "\n",
    "    image_buffer = result.segmentation_result.image_buffer\n",
    "    print(\" Segmentation result:\")\n",
    "    print(\"   Output image buffer size (bytes) = {}\".format(len(image_buffer)))\n",
    "    print(\"   Output image height = {}\".format(result.segmentation_result.image_height))\n",
    "    print(\"   Output image width = {}\".format(result.segmentation_result.image_width))\n",
    "\n",
    "    output_image_file = \"output/image_without_bg.png\"\n",
    "    with open(output_image_file, 'wb') as binary_file:\n",
    "        binary_file.write(image_buffer)\n",
    "    print(\"   File {} written to disk\".format(output_image_file));\n",
    "\n",
    "else:\n",
    "\n",
    "    error_details = sdk.ImageAnalysisErrorDetails.from_result(result)\n",
    "    print(\" Analysis failed.\")\n",
    "    print(\"   Error reason: {}\".format(error_details.reason))\n",
    "    print(\"   Error code: {}\".format(error_details.error_code))\n",
    "    print(\"   Error message: {}\".format(error_details.message))\n",
    "    print(\" Did you set the computer vision endpoint and key?\")\n",
    "\n",
    "\n",
    "# Εμφάνιση της δημιουργημένης εικόνας\n",
    "plt.imshow(Image.open(output_image_file))\n",
    "plt.axis(\"off\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "vision_source = sdk.VisionSource(filename='output/image_without_bg.png')\n",
    "analysis_options = sdk.ImageAnalysisOptions()\n",
    "\n",
    "analysis_options.features = (\n",
    "    sdk.ImageAnalysisFeature.CROP_SUGGESTIONS |\n",
    "    sdk.ImageAnalysisFeature.CAPTION |\n",
    "    sdk.ImageAnalysisFeature.DENSE_CAPTIONS |\n",
    "    sdk.ImageAnalysisFeature.OBJECTS |\n",
    "    sdk.ImageAnalysisFeature.PEOPLE |\n",
    "    sdk.ImageAnalysisFeature.TEXT |\n",
    "    sdk.ImageAnalysisFeature.TAGS\n",
    ")\n",
    "\n",
    "\n",
    "image_analyzer = sdk.ImageAnalyzer(service_options, vision_source, analysis_options)\n",
    "\n",
    "result = image_analyzer.analyze()\n",
    "\n",
    "if result.reason == sdk.ImageAnalysisResultReason.ANALYZED:\n",
    "\n",
    "    print(\" Image height: {}\".format(result.image_height))\n",
    "    print(\" Image width: {}\".format(result.image_width))\n",
    "    print(\" Model version: {}\".format(result.model_version))\n",
    "\n",
    "    if result.caption is not None:\n",
    "        print(\" \\033[32mCaption:\\033[0m\")\n",
    "        print(\"   '{}', Confidence {:.4f}\".format(result.caption.content, result.caption.confidence))\n",
    "\n",
    "    if result.dense_captions is not None:\n",
    "        print(\" \\033[32mDense Captions:\\033[0m\")\n",
    "        for caption in result.dense_captions:\n",
    "            print(\"   '{}', {}, Confidence: {:.4f}\".format(caption.content, caption.bounding_box, caption.confidence))\n",
    "\n",
    "    if result.objects is not None:\n",
    "        print(\" \\033[32mObjects:\\033[0m\")\n",
    "\n",
    "        # Εκτύπωση εικόνας\n",
    "        im = Image.open('output/image_without_bg.png')\n",
    "        fig, ax = plt.subplots(figsize=(12, 12))\n",
    "        plt.axis('off')\n",
    "        ax.imshow(im)\n",
    "\n",
    "        for object in result.objects:\n",
    "            print(\"   '{}', {}, Confidence: {:.4f}\".format(object.name, object.bounding_box, object.confidence))\n",
    "            if object.confidence > 0.3:\n",
    "                # Εκτύπωση του bounding box\n",
    "                rect = patches.Rectangle((object.bounding_box.x, object.bounding_box.y), \n",
    "                                         object.bounding_box.w, object.bounding_box.h, \n",
    "                                         linewidth=2, edgecolor='r', facecolor='none')\n",
    "                ax.add_patch(rect)\n",
    "                plt.annotate(\"{} ({:.2f}%)\".format(object.name, object.confidence * 100), \n",
    "                             (object.bounding_box.x,object.bounding_box.y), backgroundcolor='r')\n",
    "        plt.show()\n",
    "\n",
    "    if result.tags is not None:\n",
    "        print(\" \\033[32mTags:\\033[0m\")\n",
    "        for tag in result.tags:\n",
    "            print(\"   '{}', Confidence {:.4f}\".format(tag.name, tag.confidence))\n",
    "\n",
    "    if result.people is not None:\n",
    "        print(\" \\033[32mPeople:\\033[0m\")\n",
    "        for person in result.people:\n",
    "            print(\"   {}, Confidence {:.4f}\".format(person.bounding_box, person.confidence))\n",
    "\n",
    "    if result.crop_suggestions is not None:\n",
    "        print(\" \\033[32mCrop Suggestions:\\033[0m\")\n",
    "        for crop_suggestion in result.crop_suggestions:\n",
    "            print(\"   Aspect ratio {}: Crop suggestion {}\"\n",
    "                  .format(crop_suggestion.aspect_ratio, crop_suggestion.bounding_box))\n",
    "\n",
    "    if result.text is not None:\n",
    "        print(\" \\033[32mText:\\033[0m\")\n",
    "        for line in result.text.lines:\n",
    "            points_string = \"{\" + \", \".join([str(int(point)) for point in line.bounding_polygon]) + \"}\"\n",
    "            print(\"   Line: '{}', Bounding polygon {}\".format(line.content, points_string))\n",
    "            for word in line.words:\n",
    "                points_string = \"{\" + \", \".join([str(int(point)) for point in word.bounding_polygon]) + \"}\"\n",
    "                print(\"     Word: '{}', Bounding polygon {}, Confidence {:.4f}\"\n",
    "                      .format(word.content, points_string, word.confidence))\n",
    "\n",
    "    result_details = sdk.ImageAnalysisResultDetails.from_result(result)\n",
    "    print(\" Result details:\")\n",
    "    print(\"   Image ID: {}\".format(result_details.image_id))\n",
    "    print(\"   Result ID: {}\".format(result_details.result_id))\n",
    "    print(\"   Connection URL: {}\".format(result_details.connection_url))\n",
    "    print(\"   JSON result: {}\".format(result_details.json_result))\n",
    "\n",
    "else:\n",
    "\n",
    "    error_details = sdk.ImageAnalysisErrorDetails.from_result(result)\n",
    "    print(\" Analysis failed.\")\n",
    "    print(\"   Error reason: {}\".format(error_details.reason))\n",
    "    print(\"   Error code: {}\".format(error_details.error_code))\n",
    "    print(\"   Error message: {}\".format(error_details.message))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.2"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}