model.py

import base64
import os
from openai import OpenAI
from dotenv import load_dotenv

load_dotenv()
client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))

# Function to encode the image
def encode_image(image_path):
    with open(image_path, "rb") as image_file:
        return base64.b64encode(image_file.read()).decode("utf-8")


# Path to your image
image_path = "3.jpg"

# Getting the Base64 string
base64_image = encode_image(image_path)


response = client.responses.create(
    model="gpt-4o-mini",
    input=[
        {
            "role": "user",
            "content": [
                { "type": "input_text", "text": "The following image is an image of a object that was thrown in the trashcan. Can you identify which of the following classes it belongs to? The classes: recyclable, compost, landfill. Your respond should ONLY be the object detected and the class it belongs to. Nothing else and no formatting. Example: 'banana peel, compost'. Thank you" },
                {
                    "type": "input_image",
                    "image_url": f"data:image/jpeg;base64,{base64_image}",
                },
            ],
        }
    ],
)

print(response.output_text)