Spaces:
Sleeping
Sleeping
gamingflexer
commited on
Commit
·
620e74a
1
Parent(s):
a59c27e
Add OpenAIVision class for image description generation
Browse files- src/module/llm_vision.py +59 -0
src/module/llm_vision.py
ADDED
@@ -0,0 +1,59 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import base64
|
2 |
+
import requests
|
3 |
+
from config import OPENAI_API
|
4 |
+
import os
|
5 |
+
|
6 |
+
OPENAI_API = os.getenv("OPENAI_API")
|
7 |
+
|
8 |
+
"""
|
9 |
+
openai_vision = OpenAIVision(api_key)
|
10 |
+
image_path = "path_to_your_image.jpg"
|
11 |
+
prompt = ""
|
12 |
+
response = openai_vision.get_image_description(prompt,image_path)
|
13 |
+
"""
|
14 |
+
|
15 |
+
class OpenAIVision:
|
16 |
+
def __init__(self):
|
17 |
+
self.api_key = OPENAI_API
|
18 |
+
self.base_url = "https://api.openai.com/v1/chat/completions"
|
19 |
+
|
20 |
+
def __encode_image(self, image_path):
|
21 |
+
with open(image_path, "rb") as image_file:
|
22 |
+
return base64.b64encode(image_file.read()).decode('utf-8')
|
23 |
+
|
24 |
+
def get_image_description(self, image_path, prompt):
|
25 |
+
base64_image = self.__encode_image(image_path)
|
26 |
+
|
27 |
+
headers = {
|
28 |
+
"Content-Type": "application/json",
|
29 |
+
"Authorization": f"Bearer {self.api_key}"
|
30 |
+
}
|
31 |
+
|
32 |
+
payload = {
|
33 |
+
"model": "gpt-4-vision-preview",
|
34 |
+
"temperature": 0.0,
|
35 |
+
"messages": [
|
36 |
+
{
|
37 |
+
"role": "user",
|
38 |
+
"content": [
|
39 |
+
{
|
40 |
+
"type": "text",
|
41 |
+
"text": prompt,
|
42 |
+
},
|
43 |
+
{
|
44 |
+
"type": "image_url",
|
45 |
+
"image_url": {
|
46 |
+
"url": f"data:image/jpeg;base64,{base64_image}"
|
47 |
+
}
|
48 |
+
}
|
49 |
+
|
50 |
+
]
|
51 |
+
|
52 |
+
}
|
53 |
+
],
|
54 |
+
"max_tokens": 1000,
|
55 |
+
|
56 |
+
}
|
57 |
+
|
58 |
+
response = requests.post(self.base_url, headers=headers, json=payload)
|
59 |
+
return response.json()
|