import os
import torch
import gradio as gr

from model import create_effnetb2_model
from timeit import default_timer as timer
from typing import List, Dict, Tuple

class_names = ["pizza", "steak", "sushi"]

effnetb2, effnetb2_transform = create_effnetb2_model(num_classes=len(class_names))

state_dict = torch.load(
    "pretrained_effnetb2_pizza_steak_sushi_20_percent.pt",
    map_location=torch.device("cpu")  # Load model to the CPU
)

effnetb2.load_state_dict(state_dict)

def predict(img) -> Tuple[Dict, float]:
    start_time = timer()

    img = effnetb2_transform(img).unsqueeze(0)

    effnetb2.eval()
    with torch.inference_mode():
        pred_probs = torch.softmax(effnetb2(img), dim=1)

    pred_labels_and_probs = {class_names[i]: float(pred_probs[0][i]) for i in range(len(class_names))}

    end_time = timer()

    return pred_labels_and_probs, round(end_time-start_time, 4)

title = "FoodVision Mini"
description = "An [EfficientNetB2 feature extractor](www.google.com) computer vision model to classify images as pizza, steak, sushi"

example_list = [["examples/" + example] for example in os.listdir("examples")]

demo = gr.Interface(fn=predict, 
                    inputs=gr.Image(type="pil"),
                    outputs=[gr.Label(num_top_classes=3, label="Predictions"),
                            gr.Number(label="Prediction time (s)")],
                    examples=example_list,
                    title=title,
                    description=description)

demo.launch(debug=False)