app.py

import torch
from typing import Tuple, Dict
from timeit import default_timer as timer
import os
import gradio as gr

from model import create_effnetb2

with open("class_names.txt", "r") as f:
    classes = [food.strip() for food in f.readlines()]

effnetb2, effnetb2_transform = create_effnetb2(len(classes), 42)

effnetb2.load_state_dict(torch.load(f="09_pretrained_effnetb2_feature_extractor_food101_20_percent.pth", map_location=torch.device("cpu")))

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

    img = effnetb2_transform(img).unsqueeze(0)

    effnetb2.eval()
    with torch.inference_mode():
        probs = torch.softmax(effnetb2(img), dim=1)
    
    pred_label_and_probs = {classes[i]: float(probs[0][i]) for i in range(len(classes))}

    pred_time = round(timer() - start_time, 5)

    return pred_label_and_probs, pred_time

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

title = "FoodVision Big"
description = "An EffNetB2 feature extractor that characterizes images of food into one of 101 different classes"
article = "Created through Pytorch Zero To Hero"

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, article=article)

demo.launch()