A practical approach is to use a command-line downloader such as wget. It can fetch image files from regular web pages, crawl through linked page hierarchies recursively, and also works with FTP sources if a site offers them. With filtering, you can target image file types, though you may still pick up some site graphics that need to be removed afterward.

The bigger issue is verification: a site claiming images are public domain does not guarantee that every file actually is. Community experience suggests that some “free” collections include misattributed, copied, or trademarked material. So if licensing certainty matters, you should verify the source carefully rather than assuming a bulk collection is safe.

If you only need images for private testing, one answer noted that some people rely on a fair-use interpretation, but that is not a reliable substitute for confirmed rights. A safer route is to get permission directly from photographers or use sources with clearly documented licensing.