app.py

import requests
import gradio as gr
from PIL import Image
import io
import base64 
from pathlib import Path

url = 'https://source.unsplash.com/random/'

def fn():
    pass

def getJpeg(url):

    response = requests.get(url)

    response.content

    in_memory_file = io.BytesIO(response.content)
    im = Image.open(in_memory_file)
    return im

def encode_image(url):

    # To convert your image file into the base64 format required by the API
    img_b64 = gr.processing_utils.encode_url_or_file_to_base64(url)
    return url, img_b64

def get_caption(img_b64):
    
    r = requests.post(url='https://hf.space/embed/OFA-Sys/OFA-Image_Caption/+/api/predict/', json={"data": [img_b64]})
    result = r.json()
    return result['data'][0]

def fn(url):
    img_url, img_b64 = encode_image(url)
    #im = getJpeg(img_url)
    write_image(decode_image(img_b64))
    cap = get_caption(img_b64)
    return cap,Image.open('./image.png')
    

#img_url, img_b64 = encode_image(url)

def decode_image(img_b64):
    img_b64 =img_b64.partition(",")[2]
    #decode base64 string data
    decoded_data=base64.b64decode((img_b64))
    return decoded_data

def write_image(img):
    img_file = open('./image.png', 'wb')
    img_file.write(img)
    img_file.close()


'''

io = gr.Interface(
    fn= fn, inputs=gr.Textbox(url, label='Default URL'),
    outputs=[gr.Text(label='Model\'s Description'), gr.Image(label='Image')]
    
)
'''
io = gr.Interface(
    description="Machine Vision Generated Description of Image @ URL",
    fn= fn, inputs='text' ,
    outputs=['text', 'image'],
    examples=[
         ['https://source.unsplash.com/random/'],
         ['https://assets.bwbx.io/images/users/iqjWHBFdfxIU/iz_kgFAaRVSs/v1/-1x-1.jpg']
        
        ],
    article=Path("./description.md").read_text()
    )
io.launch()