JUNGU's picture
Update app.py
15448c9 verified
raw
history blame
1.97 kB
import gradio as gr
from openai import OpenAI
import os
from PIL import Image
import base64
import io
# OpenAI ํด๋ผ์ด์–ธํŠธ๋ฅผ API ํ‚ค๋กœ ์ดˆ๊ธฐํ™”( ๋ ˆ๊ฑฐ์‹œ๊ฐ€ ์˜ค๋ฅ˜๋‚˜์„œ ์ƒˆ๋กœ ์ˆ˜์ •ํ•จ)
api_key = os.getenv("OPENAI_API_KEY")
if api_key is None:
raise ValueError("OPENAI_API_KEY ํ™˜๊ฒฝ ๋ณ€์ˆ˜๊ฐ€ ์„ค์ •๋˜์ง€ ์•Š์•˜์Šต๋‹ˆ๋‹ค.")
client = OpenAI(api_key=api_key)
def image_to_base64(image):
buffered = io.BytesIO()
image.save(buffered, format="JPEG")
img_str = base64.b64encode(buffered.getvalue()).decode()
return img_str
def extract_and_summarize(image):
# Convert image to base64
image_base64 = image_to_base64(image)
# Prepare the prompt for GPT-4
prompt = [
{
"role": "system",
"content": "You are a helpful assistant. Summarize the text content of the document image provided."
},
{
"role": "user",
"content": [
{"type": "text", "text": "์—ฌ๊ธฐ ์ด๋ฏธ์ง€๋กœ ๋œ ๋ฌธ์„œ๊ฐ€ ์žˆ์–ด. ๋ฌธ์„œ๋ฅผ ๋ณด๊ณ  ๋‚ด์šฉ์„ 3์ค„๋กœ ์š”์•ฝํ•ด. ์ œ์ถœํ•ด์•ผํ•˜๋Š” ๋ฌธ์„œ์˜ ๊ฒฝ์šฐ ์ œ์ถœ ๊ธฐ๊ฐ„๊ณผ ๋ฐฉ๋ฒ•์„ ๋ฐ˜๋“œ์‹œ ํฌํ•จํ•ด"},
{"type": "image_url", "image_url": {"url": f"data:image/jpeg;base64,{image_base64}"}}
]
}
]
# Call GPT-4 API for summarization
response = client.chat.completions.create(
model="gpt-4o",
messages=prompt,
temperature=0.0,
max_tokens=300,
)
# Extract summary from GPT-4 response
summary = response.choices[0].message.content
return summary
# Define Gradio interface
iface = gr.Interface(
fn=extract_and_summarize,
inputs=gr.Image(type="pil", label="Upload Document Image"),
outputs=gr.Textbox(label="Summarized Text"),
title="๊ณต๋ฌธ์„œ ์š”์•ฝ ์ƒ์„ฑ๊ธฐ",
description="๋ฌธ์„œ์˜ ํ™”๋ฉด์„ ์บก์ณํ•˜์—ฌ ์—…๋กœ๋“œํ•˜๋ฉด ์š”์•ฝํ•ด์ค๋‹ˆ๋‹ค."
)
# Launch the interface
iface.launch()