app.py

import streamlit as st
import torch
import torch.nn as nn
from torchvision import transforms
from PIL import Image
from huggingface_hub import hf_hub_download
import os 

# Define the CNN model class
class CNNClassifier(nn.Module):
    def __init__(self, n_classes):
        super(CNNClassifier, self).__init__()
        self.model = nn.Sequential(
            nn.Conv2d(3, 64, kernel_size=3, stride=1, padding=1),
            nn.BatchNorm2d(64),
            nn.ReLU(),
            nn.MaxPool2d(2, stride=2),

            nn.Conv2d(64, 128, kernel_size=3, stride=1, padding=1),
            nn.BatchNorm2d(128),
            nn.ReLU(),
            nn.Dropout(0.2),
            nn.MaxPool2d(2, stride=2),

            nn.Conv2d(128, 256, kernel_size=3, stride=1, padding=1),
            nn.BatchNorm2d(256),
            nn.ReLU(),
            nn.MaxPool2d(2, stride=2),

            nn.AdaptiveAvgPool2d(1),
            nn.Flatten(),
            nn.Linear(256, n_classes)
        )

    def forward(self, x):
        return self.model(x)

hf_token = os.getenv("HF_TOKEN")

# Load the model from Hugging Face
model_path = hf_hub_download(repo_id="louiecerv/cats_dogs_recognition_torch_cnn", 
    filename="cats_dogs_classifier.pth", use_auth_token=hf_token)
n_classes = 2
model = CNNClassifier(n_classes)
model.load_state_dict(torch.load(model_path))
model.eval()

# Define the transformation pipeline
transform = transforms.Compose([
    transforms.Resize((128, 128)),
    transforms.ToTensor(),
])

# Streamlit app
st.title("Cat vs Dog Classifier")
st.write("Upload an image and the model will classify it as a cat or a dog.")

uploaded_file = st.file_uploader("Choose an image...", type=["jpg", "jpeg", "png"])

if uploaded_file is not None:
    image = Image.open(uploaded_file)
    st.image(image, caption="Uploaded Image", use_container_width=True)
    
    # Preprocess the image
    image = transform(image).unsqueeze(0)
    
    # Make prediction
    with torch.no_grad():
        outputs = model(image)
        _, predicted = torch.max(outputs, 1)
        label = "Cat" if predicted.item() == 0 else "Dog"
    
    st.write(f"The model predicts this image is a: **{label}**")