delafields.github.io

Company Colors 🎨

I was inspired by the research done by bold. web design on their Fortune 500 palettes site to compile a dataset around company logo/brand/website colors.

This repo contains brand palettes for said companies, palettes extracted from said companies homepages, as well as each company’s logo & a screenshot of their homepage. Do with it what you may (I have some links at the bottom for possible analysis tools).

Scraping Scripts (listed in order of execution)

• logo&palette_scraper.py: hits the bold. site, looping over various industries and scrapes company name/industy/brand palette, and logo location
• download_logos.py: downloads the logos from the last bullet into logos/
• get_urls.py: takes the company names and does a quick Google search for their homepage urls
• take_screenshots.py: pops open a headless Chrome browser and screenshots the urls from above. Saves them to a hidden screenshots/ folder. Hidden because ~see next bullet~
• bulk_resize_images.py: resizes the screenshots to 512x512 images
• extract_screenshot_colors.py: takes said screenshots and uses the colorgram package to extract the top 6 colors in the screenshot

Data

logo_colors.csv (sourced from the bold. site)

• company: company name, hypen separated
• category: industry
• color_{1-8}: contains 1-8 hex codes of brand colors (as determined by bold.)

screenshot_colors.csv (extracted from website screenshots using colorgram)

• company: company name, hypen separated
• color_{1-6}: contains 1-6 hex codes of colors in the screenshot
• color_{1-6}_proportion: proportion of the screenshot that contains said color.
  • Note this may over-index in whites or space-filler colors. [WIP]
• screenshot_location: where the screenshot is saved

company_urls.csv

• company: company name, hypen separated
• url: the company’s homepage to be screenshotted

logo_locations.csv

• company: company name, hypen separated
• file_name: where the logo is saved
• url: the url of the logo to be downloaded

Links to helpers/tutorials

• How to website screenshot: https://dev.to/bilal_io/website-screenshots-with-selenium-in-python-kcn
• Color extractor package: https://github.com/obskyr/colorgram.py
• Color namer: https://graphicdesign.stackexchange.com/questions/5120/how-can-i-get-the-closest-color-word-for-a-hex-color
• Calculating color similarity: https://stats.stackexchange.com/questions/109618/computing-image-similarity-based-on-color-distribution
• Making a color histogram: https://stackoverflow.com/questions/12182891/plot-image-color-histogram-using-matplotlib/12183468