app.py

# Transformers and its models
import transformers

# For Image Processing
from transformers import ViTImageProcessor

# For Model
from transformers import ViTModel, ViTConfig

# For data augmentation
from torchvision import transforms, datasets

# For GPU
from transformers import set_seed
from torch.optim import AdamW
from accelerate import Accelerator, notebook_launcher

# For Data Loaders
import datasets
from torch.utils.data import Dataset, DataLoader

# For Display
from tqdm.notebook import tqdm

# Other Generic Libraries
import torch
import PIL
import streamlit as st
import gc
from glob import glob
import shutil
import torch.nn.functional as F

# Initialse Globle Variables
MODEL_TRANSFORMER = 'google/vit-base-patch16-224'
BATCH_SIZE = 8

# Set the device (GPU or CPU)
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

pipeline = pipeline(task="image-classification", model="julien-c/hotdog-not-hotdog")

st.title("Hot Dog? Or Not?")

# Read images from directory


# Read image from Camera
enable = st.checkbox("Enable camera")
picture = st.camera_input("Take a picture", disabled=not enable)
if picture:
    col1, col2 = st.columns(2)
    image = PIL.Image.open(picture)
    col1.image(image, use_column_width=True)
    predictions = pipeline(image)

    col2.header("Probabilities")
    for p in predictions:
        col2.subheader(f"{ p['label'] }: { round(p['score'] * 100, 1)}%")