import requests
import gradio as gr
import torch
from timm import create_model
from timm.data import resolve_data_config
from timm.data.transforms_factory import create_transform 

# url for accesing the image DB
IMAGENET_1k_URL = "https://storage.googleapis.com/bit_models/ilsvrc2012_wordnet_lemmas.txt"
# fetching labels from the URL
LABELS = requests.get(IMAGENET_1k_URL).text.strip().split('\n')

#using a pretrained resnet50 model
model = create_model('resnet50',pretrained=True)
transform = create_transform(**resolve_data_config({},model=model))
# we do not need to train model , hence using model.eval() to use it only for inference 
model.eval()

# declaring the main fn. for returning the prediction from our model 
# we use softmax, to take probabilities of the outputs.
def predict(img):
    img = img.convert('RGB')
    img = transform(img).unsqueeze(0)
    with torch.no_grad():
        out=  model(img)
    probability = torch.nn.functional.softmax(out[0],dim=0)

    values, indices = torch.topk(probability,k=5)
    return {LABELS[i]: v.item() for i,v in zip(indices,values)}


iface = gr.Interface(fn=predict, inputs=gr.inputs.Image(type='pil'), outputs="label").launch()
iface.launch()