Image to image with Azure Computer Vision and Azure Open AI¶

import datetime
import glob
import openai
import os
import requests
import sys
import time

from dotenv import load_dotenv
from io import BytesIO
from IPython.display import Image as viewimage
from matplotlib import pyplot as plt
from PIL import Image

IMAGES_DIR = "bing_images"
RESULTS_DIR = "artificial_images"

os.makedirs(IMAGES_DIR, exist_ok=True)
os.makedirs(RESULTS_DIR, exist_ok=True)

def image_bing_search(search_term, idx=0):
    """
    Get image from Bing Search
    """
    # Get key and endpoint for the Bing Search service
    load_dotenv("azure.env")
    bing_key = os.getenv("bing_key")
    bing_search_url = os.getenv("bing_search_url")

    # Settings
    headers = {"Ocp-Apim-Subscription-Key": bing_key}
    params = {"q": search_term, "license": "public", "imageType": "photo"}

    # Running the bing search service
    response = requests.get(bing_search_url, headers=headers, params=params)

    response.raise_for_status()
    search_results = response.json()

    # Getting the image
    thumbnail_url = search_results["value"][idx]["thumbnailUrl"]
    image = requests.get(thumbnail_url)
    image.raise_for_status()
    img = Image.open(BytesIO(image.content))

    # Saving the image
    output_file = os.path.join(IMAGES_DIR, search_term + ".jpg")
    img.save(output_file)

    return img, output_file

def get_caption(image_file):
    """
    Get caption from an image using Azure Computer Vision 4
    """
    # Get key and endpoint for Azure Computer Vision service
    load_dotenv("azure.env")
    key = os.getenv("azure_cv_key")
    endpoint = os.getenv("azure_cv_endpoint")

    # settings
    options = "&features=caption,tags"
    model = "?api-version=2023-02-01-preview&modelVersion=latest"
    url = endpoint + "/computervision/imageanalysis:analyze" + model + options
    headers = {
        "Content-type": "application/octet-stream",
        "Ocp-Apim-Subscription-Key": key,
    }

    # Read the image file
    with open(image_file, "rb") as f:
        data = f.read()

    # Sending the requests
    response = requests.post(url, data=data, headers=headers)
    results = response.json()

    # Parsing the results
    image_caption = results["captionResult"]["text"]
    tags = results["tagsResult"]["values"]
    tags_string = ", ".join(item["name"] for item in tags)
    caption = image_caption + ", " + tags_string

    return caption

def dalle2(prompt, size=1024):
    """
    Generate an image from a prompt with Dall e 2
    """
    # prompt
    extra_prompt = "full view, detailled, 8K"
    image_prompt = prompt + extra_prompt

    # Get the endpoint and key for Azure Open AI
    load_dotenv("azure.env")
    openai.api_base = os.getenv("azure_openai_endpoint")
    openai.api_key = os.getenv("azure_openai_key")

    api_version = "2023-06-01-preview"
    url = (
        f"{openai.api_base}/openai/images/generations:submit?api-version={api_version}"
    )
    headers = {
        "api-key": openai.api_key,
        "Content-Type": "application/json",
    }

    if size == 512:
        imagesize = "512x512"
    elif size == 1024:
        imagesize = "1024x1024"
    elif size == 256:
        imagesize = "256x256"
    else:
        imagesize = "1024x1024"

    body = {"prompt": image_prompt, "size": imagesize, "n": 1}

    submission = requests.post(url, headers=headers, json=body)
    operation_location = submission.headers["operation-location"]
    status = ""

    while status != "succeeded":
        time.sleep(1)
        response = requests.get(operation_location, headers=headers)
        status = response.json()["status"]

    # Get the first generated image
    image_url = response.json()["result"]["data"][0]["url"]
    response = requests.get(image_url)
    img = Image.open(BytesIO(response.content))

    # Saving the artificial image
    output_file = os.path.join(RESULTS_DIR, "dalle2_" + search_term + ".jpg")
    img.save(output_file)

    return img, output_file

def display_images(imagefile1, imagefile2):
    """
    Display two images side by side
    """
    # Reading images
    img1 = plt.imread(imagefile1)
    img2 = plt.imread(imagefile2)

    # Subplots
    f, ax = plt.subplots(1, 2, figsize=(15, 8))
    ax[0].imshow(img1)
    ax[1].imshow(img2)

    # Titles
    ax[0].set_title("Bing Search image")
    ax[1].set_title("DALL·E 2 artificial image")
    title = "Image to image with Bing Search, Azure Computer Vision and Azure Open AI"
    f.suptitle(title, fontsize=15)
    plt.tight_layout()

    # Removing the axis
    ax[0].axis("off")
    ax[1].axis("off")

    plt.show()

search_term = "a portrait of a man, ages 30-40"

search_term = "a red motorcycle in a street"

search_term = "lake view from a drone"

search_term = "beautiful garden with trees and a pond"

search_term = "a mountain peak with clouds"

search_term = "a close up of purple orchids"