For effective bulk-scanning of documents, I'm using an Epson GT-S50* document scanner. This scanner connects to a host PC via a USB cable. Software provided by Epson manages the scans, producing the final output of image files or PDF multi-page documents.
Unlike the popular trend of combining multiple functions into one device, this product does only one thing - scanning documents - but it does it very well. This scanner handles separated sheets of paper (single- or double-sided), but handles up to 75 pages with its document feeder.
The efficiency of this scanner comes from its reliability and the way it operates - it scans the pages as fast as it can into its internal buffer, and while doing the scanning, begins feeding the scan to the host PC, which finishes the processing of the document. In light-duty scanners, this is done serially, one-page at a time, which really slows down scanning by requiring you to stand there waiting for each page to be completely processed.
I normally scan pages at 200 DPI, which is the minimum for character recognition by the included software. Even for a double-sided page, this takes no more than a few seconds per page, so I wait while the entire document is scanned to make sure the pages feed without a problem, then I can walk away while the scans are processed into a PDF or separate TIFF image files. Depending on the speed of the host computer and options selected, the post-processing of the scans can take a considerable amount of time, so being able to walk away right after the pages are actually scanned can save a lot of your time.
Speaking of the document feeding, this feeder works very reliably for a fast scanner. It seems to work well even on slick pages from a magazine, or small pages from a paper-back book-sized document. The only jams I've had were due to ragged edges of pages torn from a magazine. For really odd-sized scans (e.g. checks), Epson provides a folded sheet of partially translucent plastic which holds the document as it goes through the feeder.
While the pages are being scanned, the host computer will start to do the proocessing of the pages, which include optional automatic straightening of the page, background color removal, de-screening** of magazine photos, etc. for output of PDF documents. Here's a sample scan of a magazine article. I left the image adjustment controls at their default settings for a color document, but I had auto-straightening on and de-screening** selected, with the resolution set to 200DPI. Since character-recognition was done on the scanned page, you can search for words within the article (try "cassette"). This two-page, single-sided magazine page scan took ~9 seconds to feed and scan the two sheets, and another 34 seconds to produce the PDF file. The post-processing of the file takes a relatively long time, but in this case I'm using an old (and slow) netbook type computer to run the scanner.
Another sample scan is a 4-page, double-sided scan of some hand-outs from an old course I took. This took ~9 seconds for the 4 sheets to be fed through the scanner, with another 1-1/2 minutes to finish the conversion to a PDF. Note that page 5 was printed in landscape mode on the paper, but the software automatically detected this and rotated it for the PDF file. This is handy, though it doesn't always work because it depends on the software recognizing text on the page.
For photos as well as PDF documents, you can choose color or b/w, and the resolution. At least 200 DPI is required by the software for PDF character recognition. Sharp photos should be scanned at 300DPI or higher to maintain quality, but of course, the scan time per page increases accordingly.