app.py

import os
import cv2
import uuid

import gradio as gr
import numpy as np

import neovision

MARKDOWN = """
# neovision 💬 + 📸

This is a demo of neovision, a tool that allows you to chat with your webcamusinf GTP Vision. 
"""

connector = neovision.OpanAIConnector()


def save_image_to_drive(image: np.ndarray) -> str:
    image_filename = f"{uuid.uuid4()}.jpeg"
    image_directory = "data"
    os.makedirs(image_directory, exist_ok=True)
    image_path = os.path.join(image_directory, image_filename)
    cv2.imwrite(image_path, image)
    return image_path



def respond(image: np.ndarray, prompt: str, chat_history):
    image = np.fliplr(image)
    image = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
    image_path = save_image_to_drive(image)
    response = connector.simple_prompt(image=image, prompt=prompt)
    chat_history.append(((image_path,), None))
    chat_history.append((prompt, response))
    return "", chat_history


# Define the function that will process the inputs
def process(inputs):
    image, prompt, chat_history = inputs
    return respond(image, prompt, chat_history)


with gr.Blocks() as demo:
    gr.Markdown(MARKDOWN)
    with gr.Row():
        webcam = gr.Image(source="webcam", tool="select", type="numpy", streaming=True)  # Correct component for webcam
        message = gr.Textbox()
        chatbot = gr.Chatbot()
        clear_button = gr.Button("Clear")
        submit_button = gr.Button("Submit")

    submit_button.click(
        fn=process,
        inputs=[webcam, message, chatbot.chat_history],
        outputs=[message, chatbot]
    )

    clear_button.click(
        fn=lambda: ("", []),
        inputs=[],
        outputs=[message, chatbot]
    )

demo.launch(debug=False, show_error=True)