time_machine

worldLab_2026

Description

This project extracts and processes "Then vs Now" image datasets from web articles, downloads images, geocodes locations, and splits images into past and current views.

Setup

1. Install uv: curl -LsSf https://astral.sh/uv/install.sh | sh
2. Create virtual environment: uv venv
3. Activate and install dependencies: source .venv/bin/activate && uv pip install requests beautifulsoup4 pillow geopy

Usage

1. Place the web page content in dataset_assets/data/page_content.txt
2. Run the parser: cd dataset_assets/scripts && python parse_data.py
3. Download images: python downloader.py
4. Split images: python splitter.py

Dataset Structure

The processed dataset is stored in dataset_assets/data/dataset.json with the following format:

• title: Section title
• location: Extracted location
• date: Year
• image_url: URL of the image
• description: Text description
• geolocation: Latitude, longitude, and address (if available)

Split images are saved in dataset_assets/images/ and dataset_assets/split_images/.

Sample Images

Here are some examples from the dataset: