app.py

import streamlit as st
from PIL import Image
import os
import json
import numpy as np
import pickle
import zipfile
from io import BytesIO
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score

# Directories
UPLOAD_DIR = "uploaded_images"
MODEL_PATH = os.path.join(UPLOAD_DIR, "model.pkl")
LABELS_PATH = os.path.join(UPLOAD_DIR, "labels.json")
os.makedirs(UPLOAD_DIR, exist_ok=True)

# Initialize model and labels
if os.path.exists(MODEL_PATH):
    with open(MODEL_PATH, "rb") as f:
        model = pickle.load(f)
else:
    model = RandomForestClassifier()

if os.path.exists(LABELS_PATH):
    with open(LABELS_PATH, "r") as f:
        labels = json.load(f)
else:
    labels = {}

# Helper function to save uploaded images
def save_uploaded_image(uploaded_file):
    file_path = os.path.join(UPLOAD_DIR, uploaded_file.name)
    with open(file_path, "wb") as f:
        f.write(uploaded_file.getbuffer())
    return file_path

# Feature extraction function
def extract_features(file_path):
    try:
        with Image.open(file_path) as img:
            return np.array(img.resize((64, 64))).flatten()
    except Exception as e:
        print(f"Error processing image {file_path}: {e}")
        return None

# Train the model
def train_model(training_data):
    global model, labels

    features, targets = [], []
    for file_path, label in training_data.items():
        feature = extract_features(file_path)
        if feature is not None:
            features.append(feature)
            targets.append(label)

    if features:
        features = np.array(features)
        targets = np.array(targets)

        model.fit(features, targets)

        with open(MODEL_PATH, "wb") as f:
            pickle.dump(model, f)

        with open(LABELS_PATH, "w") as f:
            json.dump(labels, f)
    else:
        print("No valid features found for training.")

# Classify an image
def classify_image(file_path):
    global model
    features = extract_features(file_path)
    if features is not None:
        features = features.reshape(1, -1)
        return model.predict(features)[0]
    else:
        return "Invalid Image"

# Create a sample dataset for download
def create_sample_dataset():
    sample_dir = "sample_dataset"
    os.makedirs(sample_dir, exist_ok=True)

    # Create Human and Alien directories
    human_dir = os.path.join(sample_dir, "Human")
    alien_dir = os.path.join(sample_dir, "Alien")
    os.makedirs(human_dir, exist_ok=True)
    os.makedirs(alien_dir, exist_ok=True)

    # Add placeholder images (replace with real images in a practical application)
    for i in range(1, 4):
        human_image_path = os.path.join(human_dir, f"human_{i}.jpg")
        alien_image_path = os.path.join(alien_dir, f"alien_{i}.jpg")
        Image.new('RGB', (64, 64), color=(255, 0, 0)).save(human_image_path)
        Image.new('RGB', (64, 64), color=(0, 255, 0)).save(alien_image_path)

    # Create a ZIP file for download
    zip_buffer = BytesIO()
    with zipfile.ZipFile(zip_buffer, "w") as zip_file:
        for folder_name, subfolders, filenames in os.walk(sample_dir):
            for filename in filenames:
                file_path = os.path.join(folder_name, filename)
                arcname = os.path.relpath(file_path, sample_dir)
                zip_file.write(file_path, arcname)

    return zip_buffer.getvalue()

# Streamlit app
def main():
    st.title("Human or Alien Identification")

    st.markdown(
        """
        Welcome to the **Human or Alien Identification App**! Here's what you can do:

        - **Identify Image:** Upload an image and classify it as "Human" or "Alien." The classifications you save will be added to the training data.
        - **Train Model:** Review and manage the images already classified as "Human" or "Alien." Upload additional images to improve the training dataset.
        - **Download Sample Dataset:** Download a pre-structured dataset to use for training and classification.
        """
    )

    tab1, tab2, tab3 = st.tabs(["Identify Image", "Train Model", "Download Sample Dataset"])

    with tab1:
        st.header("Identify Image")
        uploaded_files = st.file_uploader("Upload Images to Identify", type=["jpg", "jpeg", "png"], accept_multiple_files=True)

        if uploaded_files:
            results = {}

            for uploaded_file in uploaded_files:
                try:
                    file_path = save_uploaded_image(uploaded_file)
                    st.image(file_path, caption=f"Uploaded Image: {uploaded_file.name}", use_column_width=True)

                    prediction = classify_image(file_path)
                    results[file_path] = prediction
                except Exception as e:
                    st.error(f"Error processing file {uploaded_file.name}: {e}")

            st.subheader("Classification Results")
            for file_path, label in results.items():
                st.markdown(
                    f"<p style='font-size:20px; color:blue; background-color:lightyellow; padding:10px; border-radius:5px;'>Image: {os.path.basename(file_path)} - Classified as: {label}</p>",
                    unsafe_allow_html=True,
                )

    with tab2:
        st.header("Train Model")

        st.subheader("Upload Images for Training")
        training_files = st.file_uploader("Upload Training Images", type=["jpg", "jpeg", "png"], accept_multiple_files=True)

        if training_files:
            training_data = {}

            for training_file in training_files:
                try:
                    file_path = save_uploaded_image(training_file)
                    st.image(file_path, caption=f"Uploaded Training Image: {training_file.name}", use_column_width=True)

                    label = st.radio(
                        f"Classify {training_file.name}", ["Human", "Alien"], index=0, key=training_file.name
                    )
                    training_data[file_path] = label
                except Exception as e:
                    st.error(f"Error processing file {training_file.name}: {e}")

            if st.button("Save Training Data"):
                labels.update(training_data)
                train_model(training_data)
                st.success("Training data has been saved and the model updated.")

        st.subheader("Bulk Train with Existing Dataset")
        dataset_file = st.file_uploader("Upload a JSON file containing labeled data", type=["json"])

        if dataset_file:
            dataset = json.load(dataset_file)
            training_data = {}

            for label, image_paths in dataset.items():
                for image_path in image_paths:
                    if os.path.exists(image_path):
                        training_data[image_path] = label

            if st.button("Train Model with Dataset"):
                labels.update(training_data)
                train_model(training_data)
                st.success("Model has been trained with the uploaded dataset.")

        subtab1, subtab2 = st.tabs(["Humans", "Aliens"])

        with subtab1:
            st.subheader("Human Images")
            human_images = [
                img for img, lbl in labels.items() if lbl == "Human"
            ]

            if human_images:
                st.info("These images have already been trained.")
                for image_path in human_images:
                    if os.path.exists(image_path):
                        st.image(image_path, caption=f"Image: {os.path.basename(image_path)}", use_column_width=True)
                    else:
                        st.error(f"File not found: {os.path.basename(image_path)}")
            else:
                st.warning("No human images found for training.")

        with subtab2:
            st.subheader("Alien Images")
            alien_images = [
                img for img, lbl in labels.items() if lbl == "Alien"
            ]

            if alien_images:
                st.info("These images have already been trained.")
                for image_path in alien_images:
                    if os.path.exists(image_path):
                        st.image(image_path, caption=f"Image: {os.path.basename(image_path)}", use_column_width=True)
                    else:
                        st.error(f"File not found: {os.path.basename(image_path)}")
            else:
                st.warning("No alien images found for training.")

    with tab3:
        st.header("Download Sample Dataset")

        if st.button("Download Sample Dataset"):
            sample_dataset = create_sample_dataset()
            st.download_button(
                label="Click to Download",
                data=sample_dataset,
                file_name="sample_dataset.zip",
                mime="application/zip"
            )

if __name__ == "__main__":
    main()