The following is one of the techniques to detect objects in images 

#! /usr/bin/python

import requests

import cv2

import numpy as np

from sklearn.cluster import DBSCAN

from sklearn.preprocessing import normalize

import hdbscan

import matplotlib.pyplot as plt

from io import BytesIO

from azure.core.credentials import AzureKeyCredential

from azure.ai.vision.imageanalysis import ImageAnalysisClient

from azure.ai.vision.imageanalysis.models import VisualFeatures

import os

match_threshold = 0.5

min_number_of_cluster_members = 2

object_uri = os.getenv("AZURE_RED_CAR_2_SAS_URL").strip('"')

scene_uri = os.getenv("AZURE_QUERY_SAS_URI").strip('"')

# Step 1: Download images from SAS URLs

def download_image(url):

    response = requests.get(url)

    image_array = np.frombuffer(response.content, np.uint8)

    return cv2.imdecode(image_array, cv2.IMREAD_COLOR)

# Step 2: Use OpenCV template matching to find object occurrences

def count_object_occurrences(scene, template, threshold=match_threshold):

    scene_gray = cv2.cvtColor(scene, cv2.COLOR_BGR2GRAY)

    template_gray = cv2.cvtColor(template, cv2.COLOR_BGR2GRAY)

    result = cv2.matchTemplate(scene_gray, template_gray, cv2.TM_CCOEFF_NORMED)

    locations = np.where(result >= threshold)

    w, h = template_gray.shape[::-1]

    rects = [[pt[0], pt[1], pt[0] + w, pt[1] + h] for pt in zip(*locations[::-1])]

    rects, _ = cv2.groupRectangles(rects, groupThreshold=1, eps=0.5)

    return len(rects)

# Step 3: Count matches

def count_matches():

    scene_img = download_image(scene_uri)

    object_img = download_image(object_uri)

    count = count_object_occurrences(scene_img, object_img)

    return count

# print(f"Detected {count_matches()} occurrences of the object.")

#Output: Detected 1 occurrences of the object.

#Codingexercise: codingexercise-07-06-2025.pdf

https://1drv.ms/b/c/d609fb70e39b65c8/EQti1RQIDMpNrxM0jjPPYzkBI0PX5cBV-eUmQR_Js0nsBQ?e=sssW8n