That seems like a very expensive way to crawl the internet
Scrape normally collect emails, if no email seen take screenshot and OCR OCR is cheap and REGEX is cheap