display_dataset.py

# use in a ipynb notebook

import os
from PIL import Image
from IPython.core.display import HTML, display as disp
import io
import base64

# Disable truncating of text
pd.set_option('display.max_colwidth', None)

def display_images(root_dir, csv_filepath):
    # Read csv file into a pandas DataFrame
    df = pd.read_csv(csv_filepath)

    for _, row in df[:5].iterrows():
        filename = row['filename']
        if not os.path.isabs(filename):
            image_filepath = os.path.join(root_dir, filename)
        else:
            image_filepath = filename

        if os.path.exists(image_filepath) and image_filepath.lower().endswith(('.png', '.jpg', '.jpeg', '.tiff', '.bmp', '.gif')):
            try:
                # Open image with PIL
                pil_im = Image.open(image_filepath)
                # Convert PIL image to base64-encoded image
                image_data = io.BytesIO()
                pil_im.save(image_data, format='PNG')
                base64_image = base64.b64encode(image_data.getvalue()).decode('utf-8')

                # Display image and text
                image_html = f"<img style='width: 400px; margin:0px; float: left; border: 1px solid black;' src='data:image/png;base64,{base64_image}' />"
                data_html = f"<div style='margin-left: 420px; border: 1px solid black;'>{row.to_frame().to_html()}</div>"
                final_html = image_html + data_html

                disp(HTML(final_html))

            except Exception as e:
                print(f"Failed to display image {image_filepath}. Error: {e}")
        else:
            print(f"File {image_filepath} does not exist or is not a valid image file")

# Usage
root_dir = r"D:\Andrew\Downloads\demo"
csv_filepath = r"D:\Andrew\Downloads\demo\metrics.csv"
display_images(root_dir, csv_filepath)