import torch
import os
import random
import gradio as gr
from transformers import pipeline
import base64
from datasets import load_dataset
from diffusers import DiffusionPipeline
from huggingface_hub import login
import numpy as np

def guessanImage(model, image):
    #model = "microsoft/resnet-50"
    # st.write("using model:"+model)
    
    imgclassifier  = pipeline("image-classification", model=model)
    if image is not None:  
        description = imgclassifier(image)
    return description

with gr.Blocks() as demo:
    with gr.Tab("Lion"):
       radio = gr.Radio(["microsoft/resnet-50", "google/vit-base-patch16-224", "apple/mobilevit-small"], label="Select a Classifier", info="Image Classifier?")
        
    with gr.Tab("Tiger"):
    radio = gr.Radio(["zzmicrosoft/resnet-50", "google/vit-base-patch16-224", "apple/mobilevit-small"], label="Select a Classifier", info="Image Classifier?")



demo = gr.Interface(
    fn=guessanImage,
    inputs=[radio, gr.Image(type="pil")],
    outputs=["text"],
)

demo.launch()