app.py

import streamlit as st
from datasets import load_dataset
import numpy as np
import matplotlib.pyplot as plt
import pandas as pd
import seaborn as sns

# Load dataset from Hugging Face with caching
def load_hf_dataset():
    repo_id = "louiecerv/cats_dogs_dataset"
    return load_dataset(repo_id)

dataset = load_hf_dataset()
split = "train"
data = dataset[split]
label_names = dataset[split].features["label"].names

@st.cache_data
def get_label_counts():
    return pd.Series(data["label"]).value_counts().sort_index()

def display_images(images, labels, label_names, cols=5):
    """Display images in a grid."""
    rows = (len(images) + cols - 1) // cols
    fig, axes = plt.subplots(rows, cols, figsize=(2 * cols, 2 * rows))
    axes = axes.flatten() if isinstance(axes, np.ndarray) else [axes]
    
    for i, ax in enumerate(axes):
        if i < len(images):
            ax.imshow(images[i])
            ax.set_title(label_names[labels[i]])
            ax.axis("off")
        else:
            ax.axis("off")
    
    st.pyplot(fig)

def main():
    st.title("Image Dataset Explorer")
    st.subheader(f"Displaying images from the {split} set")
    
    # Show Initial Images
    if st.button("Show First 25 Images"):
        with st.spinner("Loading images..."):
            images = [data[i]["image"] for i in range(25)]
            labels = [data[i]["label"] for i in range(25)]
            display_images(images, labels, label_names)

    st.sidebar.title("Explore the Dataset")

    # Random Image Viewer
    if st.sidebar.button("Show Random Images"):
        with st.spinner("Loading images..."):
            rand_indices = [int(i) for i in np.random.choice(len(data), 25, replace=False)]
            images = [data[i]["image"] for i in rand_indices]
            labels = [data[i]["label"] for i in rand_indices]
            display_images(images, labels, label_names)

    # Class Distribution
    if st.sidebar.button("Show Class Distribution"):
        label_counts = get_label_counts()
        label_names_map = {i: name for i, name in enumerate(label_names)}
        
        fig, ax = plt.subplots(figsize=(8, 4))
        sns.barplot(x=[label_names_map[i] for i in label_counts.index], y=label_counts.values, ax=ax)
        ax.set_title("Class Distribution")
        ax.set_ylabel("Count")
        ax.set_xlabel("Class")
        st.pyplot(fig)

    # Filter by class label
    selected_label = st.sidebar.selectbox("Filter by Label", label_names)
    if st.sidebar.button("Show Filtered Images"):
        with st.spinner("Loading images..."):
            filtered_indices = [int(i) for i in np.where(np.array(data["label"]) == label_names.index(selected_label))[0]]
            if len(filtered_indices) > 0:
                images = [data[i]["image"] for i in filtered_indices]
                labels = [data[i]["label"] for i in filtered_indices]
                display_images(images, labels, label_names)
            else:
                st.write(f"No images found for label: {selected_label}")

if __name__ == "__main__":
    main()