If you've scanned several documents in one pass (or you have a PDF where each page holds two or three receipts), you don't need to crop them apart by hand. Drag the scans into AutoCropper and it detects every document on the page, crops each one, straightens the crooked ones, and exports them as individual full-quality files. It runs in your browser on any computer, it's free for up to 5 images a day (PDFs up to 5 pages), and your full-resolution scans never leave your device, which matters when the paper in question is financial.
Paperwork piles up in batches, so scan it in batches
The one-document-per-scan workflow is how a twenty-minute job becomes an afternoon. Almost everything paper arrives in stacks:
- Receipts for an expense report or tax season: six or eight fit on the scanner glass at once.
- Letters and cards worth keeping: two or three per pass.
- Certificates, old records, warranties: the folder you've been meaning to digitize for years.
- Recipe cards and index cards: a dozen passes instead of a hundred.
Lay as many on the bed as fit with a little space between them, scan the whole thing, repeat. The splitting is the part software should do, and it's the same detection that splits multiple photos per scan: documents are just rectangles with worse handwriting.
How to batch crop scanned documents
- Scan a bedful at a time. Leave a finger's width between items and don't let them overlap. Any flatbed or all-in-one printer works; my scanning tips cover the settings, and 300 DPI is plenty for paperwork.
- Upload to AutoCropper. JPEG, PNG, TIFF, HEIC, or a multi-page PDF: a PDF is unpacked page by page and every document on every page gets detected.
- Review the detections. Each document gets a crop box; tilted ones are measured and straightened automatically. Paper is harder than photos here, and it's worth being honest about why: a white receipt on a white scanner background has faint edges, so once in a while a detection needs you. Drag its corners to fix it, or draw a box by hand. The review step exists precisely because no detector is perfect on white-on-white.
- Export. Every document saves as its own file at full scan resolution, auto-named after the source (taxes2025_1.jpeg, taxes2025_2.jpeg and so on). For receipts, it's worth opening a crop's editor and setting its date field: the date is written into the file's metadata, so your receipts sort by when they happened instead of when you scanned them.
This is the opposite job from a phone scanner app
Phone scanner apps (Adobe Scan, CamScanner, Office Lens and friends) solve a different problem: capture one page at a time with the camera, perspective-correct it, and often OCR it into a searchable PDF. Great for a contract you're signing today. But when there's a stack, page-at-a-time capture is the bottleneck, and a flatbed pass with automatic splitting runs through a shoebox of receipts while the app is still asking you to hold steady. If you need text extraction afterward, run the split files through your OCR tool of choice: one clean, straightened document per image is exactly what OCR engines like, much more than a five-receipt collage.
Ways to split a stack of scanned documents, compared
| Route | Cost | How it goes |
|---|---|---|
| Crop by hand (Preview, Paint, Photos) | Free | A minute or so per document, every document, forever |
| Photoshop's Crop and Straighten | Subscription | Batch-capable, but struggles with faint edges and gives you no review step |
| Phone scanner app | Free to subscription | One page per shot, OCR built in, phone-camera capture |
| AutoCropper | Free for 5 images/day; unlimited from $7/mo | Whole bedfuls or PDFs at once, detection plus a review step, full-resolution export, no install |
What it doesn't do
Two honest limits. AutoCropper does not OCR: you get perfectly cropped images, not searchable text, so pair it with an OCR tool if you need that. And it can't restore what the paper lost: a thermal receipt that faded to gray scans as a faded receipt. Crop the stack the day the drawer gets emptied, not five years later.
Is it safe for financial documents?
Fair question for receipts and records. Your full-resolution scan is never uploaded: it stays in your browser, and what's sent to the detection server is a heavily compressed, metadata-stripped copy (around 5% of the file size) used only to locate the document boundaries. The actual cropping happens on your machine. You can inspect exactly what's sent, request by request. I documented it with my own scans.
Frequently asked questions
Can it split a PDF of scanned documents into separate files?
Yes. Upload the PDF and every page is unpacked; each document found on each page becomes its own image file. The free tier takes PDFs up to 5 pages; paid removes the cap.
Does it OCR the documents?
No. The output is clean individual images, which happen to be the best possible input for whatever OCR tool you already use: one document per image, straightened, no neighbors.
How many documents fit in one scan?
Whatever fits with a small gap between items and no overlaps: typically six to eight receipts, or two or three letter-size pages' worth of smaller items, per pass on a letter-size flatbed.
What if a detection misses a faint receipt?
Fix it in the review step: drag the box's corners, or draw a new crop by hand. White paper on a white background is the hardest case for any detector, which is why exports always go through a review screen instead of straight to disk.
What comes out the other end?
JPEG or PNG files at the original scan's full resolution, auto-named after the source scan, with optional per-document title, description, and date written into the metadata. Ordinary files in a folder you own: drop them into your expense tool, tax folder, or family archive.