Spaces:
Sleeping
Sleeping
Create desc.py
Browse files
desc.py
ADDED
|
@@ -0,0 +1,41 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import openai
|
| 2 |
+
import base64
|
| 3 |
+
|
| 4 |
+
def describe_image_with_gpt4o(image_path, api_key):
|
| 5 |
+
"""
|
| 6 |
+
Given an image file, sends it to GPT-4o Vision API and gets a detailed description.
|
| 7 |
+
Returns the description as a string.
|
| 8 |
+
"""
|
| 9 |
+
openai.api_key = api_key
|
| 10 |
+
|
| 11 |
+
# Read and encode image
|
| 12 |
+
with open(image_path, "rb") as img_file:
|
| 13 |
+
img_base64 = base64.b64encode(img_file.read()).decode("utf-8")
|
| 14 |
+
img_data_url = f"data:image/png;base64,{img_base64}"
|
| 15 |
+
|
| 16 |
+
# Compose the prompt
|
| 17 |
+
prompt = (
|
| 18 |
+
"Describe this face with as much detail as possible, "
|
| 19 |
+
"focusing on facial features, hairstyle, expression, accessories, and notable traits. "
|
| 20 |
+
"Be specific, provide detail on Clothing, style and background"
|
| 21 |
+
"Be specific, as if explaining to an artist who will draw a sticker of this person."
|
| 22 |
+
)
|
| 23 |
+
|
| 24 |
+
# Call GPT-4o Vision API
|
| 25 |
+
response = openai.chat.completions.create(
|
| 26 |
+
model="gpt-4o",
|
| 27 |
+
messages=[
|
| 28 |
+
{
|
| 29 |
+
"role": "user",
|
| 30 |
+
"content": [
|
| 31 |
+
{"type": "text", "text": prompt},
|
| 32 |
+
{"type": "image_url", "image_url": img_data_url},
|
| 33 |
+
],
|
| 34 |
+
}
|
| 35 |
+
],
|
| 36 |
+
max_tokens=256,
|
| 37 |
+
)
|
| 38 |
+
|
| 39 |
+
# Extract the description
|
| 40 |
+
description = response.choices[0].message.content.strip()
|
| 41 |
+
return description
|