├── Object Recognition for Pre-trained model.py
└── README.md


/Object Recognition for Pre-trained model.py:
--------------------------------------------------------------------------------
 1 | from PIL import Image
 2 | import numpy as np
 3 | 
 4 | def load_image(image_path):
 5 |     return Image.open(image_path).convert('L')  # Convert to grayscale
 6 | 
 7 | def image_to_array(image):
 8 |     return np.array(image)
 9 | 
10 | def normalize_image(image_array):
11 |     mean = np.mean(image_array)
12 |     std = np.std(image_array)
13 |     return (image_array - mean) / std
14 | 
15 | def save_sub_image(image_array, top_left, size, file_name):
16 |     y, x = top_left
17 |     h, w = size
18 |     sub_image = image_array[y:y+h, x:x+w]
19 |     Image.fromarray(sub_image).save(file_name)
20 | 
21 | def match_template(image, template):
22 |     image_array = normalize_image(image_to_array(image))
23 |     template_array = normalize_image(image_to_array(template))
24 |     
25 |     img_h, img_w = image_array.shape
26 |     tmpl_h, tmpl_w = template_array.shape
27 |     
28 |     best_match = None
29 |     best_score = -float('inf')  # For NCC, higher values are better
30 |     
31 |     template_mean = np.mean(template_array)
32 |     template_std = np.std(template_array)
33 |     
34 |     for y in range(img_h - tmpl_h + 1):
35 |         for x in range(img_w - tmpl_w + 1):
36 |             sub_image = image_array[y:y+tmpl_h, x:x+tmpl_w]
37 |             sub_image_mean = np.mean(sub_image)
38 |             sub_image_std = np.std(sub_image)
39 |             
40 |             norm_sub_image = (sub_image - sub_image_mean) / sub_image_std
41 |             norm_template = (template_array - template_mean) / template_std
42 |             
43 |             score = np.sum(norm_sub_image * norm_template)
44 |             print(f"Matching region at ({x}, {y}) with score: {score}")
45 |             
46 |             # Save the best matching region for visualization
47 |             if score > best_score:
48 |                 best_score = score
49 |                 best_match = (x, y)
50 |                 save_sub_image(image_array, (y, x), (tmpl_h, tmpl_w), 'best_match_region.jpg')
51 |     
52 |     return best_match, best_score
53 | 
54 | # Provide the paths to your image and template
55 | image_path = "C:\\Users\\Lenovo\\Downloads\\download.jpeg"
56 | template_path = "C:\\Users\\Lenovo\\Downloads\\download (1).jpeg"
57 | 
58 | # Load the image and template
59 | image = load_image(image_path)
60 | template = load_image(template_path)
61 | 
62 | # Perform template matching
63 | best_match, best_score = match_template(image, template)
64 | 
65 | # Display the result
66 | if best_match:
67 |     print(f"Best match found at: {best_match} with score: {best_score}")
68 | else:
69 |     print("No match found")
70 | print(f"Image size: {image.size}")
71 | print(f"Template size: {template.size}")
72 | 
73 | 
74 | 
75 | 
76 | 
77 | 
78 | output:
79 | print(f"Image size: {image.size}")
80 | print(f"Template size: {template.size}")
81 | 
82 | 
83 | 


--------------------------------------------------------------------------------
/README.md:
--------------------------------------------------------------------------------
1 | # Object-recognition-for-pre-trained-model


--------------------------------------------------------------------------------