File size: 6,485 Bytes
91219fd
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from transformers import pipeline\n",
    "\n",
    "checkpoint = \"openai/clip-vit-large-patch14\"\n",
    "detector = pipeline(model=checkpoint, task=\"zero-shot-image-classification\")\n",
    "#checkpoint = \"google/siglip-so400m-patch14-384\"\n",
    "#detector = pipeline(task=\"zero-shot-image-classification\", model=\"google/siglip-so400m-patch14-384\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from datasets import load_dataset\n",
    "\n",
    "dataset = load_dataset('pcuenq/oxford-pets')\n",
    "dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset['train'][0]['image']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from PIL import Image\n",
    "import io\n",
    "from tqdm import tqdm\n",
    "\n",
    "labels_oxford_pets = ['Siamese', 'Birman', 'shiba inu', 'staffordshire bull terrier', 'basset hound', 'Bombay', 'japanese chin', 'chihuahua', 'german shorthaired', 'pomeranian', 'beagle', 'english cocker spaniel', 'american pit bull terrier', 'Ragdoll', 'Persian', 'Egyptian Mau', 'miniature pinscher', 'Sphynx', 'Maine Coon', 'keeshond', 'yorkshire terrier', 'havanese', 'leonberger', 'wheaten terrier', 'american bulldog', 'english setter', 'boxer', 'newfoundland', 'Bengal', 'samoyed', 'British Shorthair', 'great pyrenees', 'Abyssinian', 'pug', 'saint bernard', 'Russian Blue', 'scottish terrier']\n",
    "\n",
    "# List to store true labels and predicted labels\n",
    "true_labels = []\n",
    "predicted_labels = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "for i in tqdm(range(len(dataset['train']))):\n",
    "    # Get the image bytes from the dataset\n",
    "    image_bytes = dataset['train'][i]['image']['bytes']\n",
    "    \n",
    "    # Convert the bytes to a PIL image\n",
    "    image = Image.open(io.BytesIO(image_bytes))\n",
    "    \n",
    "    # Run the detector on the image with the provided labels\n",
    "    results = detector(image, candidate_labels=labels_oxford_pets)\n",
    "    # Sort the results by score in descending order\n",
    "    sorted_results = sorted(results, key=lambda x: x['score'], reverse=True)\n",
    "    \n",
    "    # Get the top predicted label\n",
    "    predicted_label = sorted_results[0]['label']\n",
    "    \n",
    "    # Append the true and predicted labels to the respective lists\n",
    "    true_labels.append(dataset['train'][i]['label'])\n",
    "    predicted_labels.append(predicted_label)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import accuracy_score, precision_score, recall_score\n",
    "\n",
    "# Calculate accuracy\n",
    "accuracy = accuracy_score(true_labels, predicted_labels)\n",
    "\n",
    "# Calculate precision and recall\n",
    "precision = precision_score(true_labels, predicted_labels, average='weighted', labels=labels_oxford_pets)\n",
    "recall = recall_score(true_labels, predicted_labels, average='weighted', labels=labels_oxford_pets)\n",
    "\n",
    "# Print the results\n",
    "print(f\"Accuracy: {accuracy:.4f}\")\n",
    "print(f\"Precision: {precision:.4f}\")\n",
    "print(f\"Recall: {recall:.4f}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Gradio example"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import gradio as gr\n",
    "from transformers import pipeline\n",
    "\n",
    "# Load models\n",
    "vit_classifier = pipeline(\"image-classification\", model=\"kuhs/vit-base-oxford-iiit-pets\")\n",
    "clip_detector = pipeline(model=\"openai/clip-vit-large-patch14\", task=\"zero-shot-image-classification\")\n",
    "\n",
    "labels_oxford_pets = [\n",
    "    'Siamese', 'Birman', 'shiba inu', 'staffordshire bull terrier', 'basset hound', 'Bombay', 'japanese chin',\n",
    "    'chihuahua', 'german shorthaired', 'pomeranian', 'beagle', 'english cocker spaniel', 'american pit bull terrier',\n",
    "    'Ragdoll', 'Persian', 'Egyptian Mau', 'miniature pinscher', 'Sphynx', 'Maine Coon', 'keeshond', 'yorkshire terrier',\n",
    "    'havanese', 'leonberger', 'wheaten terrier', 'american bulldog', 'english setter', 'boxer', 'newfoundland', 'Bengal',\n",
    "    'samoyed', 'British Shorthair', 'great pyrenees', 'Abyssinian', 'pug', 'saint bernard', 'Russian Blue', 'scottish terrier'\n",
    "]\n",
    "\n",
    "def classify_pet(image):\n",
    "    vit_results = vit_classifier(image)\n",
    "    vit_output = {result['label']: result['score'] for result in vit_results}\n",
    "    \n",
    "    clip_results = clip_detector(image, candidate_labels=labels_oxford_pets)\n",
    "    clip_output = {result['label']: result['score'] for result in clip_results}\n",
    "    \n",
    "    return {\"ViT Classification\": vit_output, \"CLIP Zero-Shot Classification\": clip_output}\n",
    "\n",
    "example_images = [\n",
    "    [\"example_images/dog1.jpeg\"],\n",
    "    [\"example_images/dog2.jpeg\"],\n",
    "    [\"example_images/leonberger.jpg\"],\n",
    "    [\"example_images/snow_leopard.jpeg\"],\n",
    "    [\"example_images/cat.jpg\"]\n",
    "]\n",
    "\n",
    "iface = gr.Interface(\n",
    "    fn=classify_pet,\n",
    "    inputs=gr.Image(type=\"filepath\"),\n",
    "    outputs=gr.JSON(),\n",
    "    title=\"Pet Classification Comparison\",\n",
    "    description=\"Upload an image of a pet, and compare results from a trained ViT model and a zero-shot CLIP model.\",\n",
    "    examples=example_images\n",
    ")\n",
    "\n",
    "iface.launch()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": ".venv",
   "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": "3.13.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}