app.py

import io
import os
import requests
from PIL import Image
from model import get_caption_model, generate_caption
import gradio as gr

def get_model():
    return get_caption_model()

caption_model = get_model()


def predict():
    captions = []
    pred_caption = generate_caption('tmp.jpg', caption_model)

    
    captions.append(pred_caption)

    for _ in range(4):
        pred_caption = generate_caption('tmp.jpg', caption_model, add_noise=True)
        if pred_caption not in captions:
            captions.append(pred_caption)
    
    #finalc = ' '.join([str(elem) for elem in captions])
    return captions;
def launch(inputs):
    img = Image.open(requests.get(inputs, stream=True).raw)
    img = img.convert('RGB')
    img.save('tmp.jpg')
    o=predict()
    str1=""
    for ele in o:
        str1 += "\n"+ele
    os.remove('tmp.jpg')
    return str1
iface = gr.Interface(launch, inputs="text", outputs="text",)
iface.launch(debug=True,)