File size: 3,797 Bytes
0e0b22e 8f10c38 0e0b22e 78625da 0e0b22e 339da97 0e0b22e b45fd05 0e0b22e 8f10c38 0e0b22e 8f06c65 0e0b22e 2c334ee 0e0b22e |
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 |
import base64
import os
import uuid
import time
import cv2
import gradio as gr
import numpy as np
import requests
MARKDOWN = """
# ♠ Alyxsissy GPT-Vision Testing 👁🗨
Skeleton API structure for GPT-4-Vision. Implimentation into SissyGPT WiP.
"""
AVATARS = (
"https://raw.githubusercontent.com/Nymbo/Web-Hosted-Images/main/Alyxsissy%20Favicon.png?token=GHSAT0AAAAAACJF7K6DI4PXYMGQU3TPEZ2GZL2LFXA",
"https://media.roboflow.com/spaces/openai-white-logomark.png"
)
IMAGE_CACHE_DIRECTORY = "data"
API_URL = "https://api.openai.com/v1/chat/completions"
def preprocess_image(image: np.ndarray) -> np.ndarray:
image = np.fliplr(image)
return cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
def encode_image_to_base64(image: np.ndarray) -> str:
success, buffer = cv2.imencode('.jpg', image)
if not success:
raise ValueError("Could not encode image to JPEG format.")
encoded_image = base64.b64encode(buffer).decode('utf-8')
return encoded_image
def compose_payload(image: np.ndarray, prompt: str) -> dict:
base64_image = encode_image_to_base64(image)
return {
"model": "gpt-4-vision-preview",
"messages": [
{
"role": "user",
"content": [
{
"type": "text",
"text": prompt
},
{
"type": "image_url",
"image_url": {
"url": f"data:image/jpeg;base64,{base64_image}"
}
}
]
}
],
"max_tokens": 300
}
def compose_headers(api_key: str) -> dict:
return {
"Content-Type": "application/json",
"Authorization": f"Bearer {api_key}"
}
def prompt_image(api_key: str, image: np.ndarray, prompt: str) -> str:
headers = compose_headers(api_key=api_key)
payload = compose_payload(image=image, prompt=prompt)
response = requests.post(url=API_URL, headers=headers, json=payload).json()
if 'error' in response:
raise ValueError(response['error']['message'])
return response['choices'][0]['message']['content']
def cache_image(image: np.ndarray) -> str:
image_filename = f"{uuid.uuid4()}.jpeg"
os.makedirs(IMAGE_CACHE_DIRECTORY, exist_ok=True)
image_path = os.path.join(IMAGE_CACHE_DIRECTORY, image_filename)
cv2.imwrite(image_path, image)
return image_path
def respond(api_key: str, image: np.ndarray, prompt: str, chat_history):
if not api_key:
raise ValueError(
"API_KEY is not set. "
"Please follow the instructions in the README to set it up.")
image = preprocess_image(image=image)
cached_image_path = cache_image(image)
response = prompt_image(api_key=api_key, image=image, prompt=prompt)
chat_history.append(((cached_image_path,), None))
chat_history.append((prompt, response))
return "", chat_history
with gr.Blocks(theme=gr.themes.Default(primary_hue="pink", secondary_hue="pink")) as demo:
gr.Markdown(MARKDOWN)
with gr.Row():
webcam = gr.Webcam(label="Webcam", source="webcam", streaming=True)
with gr.Column():
api_key_textbox = gr.Textbox(
label="OpenAI API KEY", type="password")
chatbot = gr.Chatbot(
height=500, bubble_full_width=False, avatar_images=AVATARS)
message_textbox = gr.Textbox()
clear_button = gr.ClearButton(message_textbox, chatbot, variant="primary")
message_textbox.submit(
fn=respond,
inputs=[api_key_textbox, webcam, message_textbox, chatbot],
outputs=[message_textbox, chatbot]
)
demo.launch(debug=False, show_error=True)
|