The reason for the difference in those two numbers is because Estimated pages  are a rough list of ALL links we find as we are crawling.
​
However, a lot of these links might be disregarded as redirections, invalid formats, links to docs ( PDFs, CSVs ) etc.
​
Therefore when we start generating the sitemap/screenshots, the crawler will only take a screenshot of only valid html  pages and ignore the rest.
​