My idea is to use Tesseract to do a really quick OCR scan of a photo, then present the OCR text and a rescan/email button choice which will email to a single preset email address. The email body will include whatever was found in OCR and attach a copy of the photo.
I know this doesn’t seem like an incredible application, but the function I am looking to fill is package tracking in warehouses/mailrooms. A truck pulls up with a manifest, receiver takes a picture of it, which is sent to the front office who will process it. The main deal is to not make it do everything, in fact to make it do only two things, be fast and be simple, quality isn’t too important. No offense to them, but most warehouse people aren’t tech oriented and don’t want to be fiddling around with lots of options and typing package info on a phone. They just want to snap a pic and hit send so they can move onto the next package. As long as the picture is somewhat clear, the OCR is just for basic text searches and a quick guess as to who to notify about the delivery.
There are big complex packages that do this and everything else but cost huge amounts of money. Since this isn’t doing any complex field mapping or error checking it should be really simple to write. Really: Picture>Tesseract>Email>Done. (I know sounds simple and is simple are a world apart.) But really this almost seems like child's play to anyone who has used Tesseract already.
If adding bar code scanning isn't hard adding that would be cool, include them as text in the email body (aka BarCode1=1z234..). Also making the rescan/email choice a multiple choice rescan/email1/email2/email3 could also be a simple upgrade. But like I said simple is better.
Make it have a demo that sends up to 10 emails for free and charge $3-5 for the full version. As a business cost it is nothing, and compared to $10k for real package tracking it is less than nothing. Just set the expectation that the OCR is going to suck (because it always does) and people won't complain about it.
I know this doesn’t seem like an incredible application, but the function I am looking to fill is package tracking in warehouses/mailrooms. A truck pulls up with a manifest, receiver takes a picture of it, which is sent to the front office who will process it. The main deal is to not make it do everything, in fact to make it do only two things, be fast and be simple, quality isn’t too important. No offense to them, but most warehouse people aren’t tech oriented and don’t want to be fiddling around with lots of options and typing package info on a phone. They just want to snap a pic and hit send so they can move onto the next package. As long as the picture is somewhat clear, the OCR is just for basic text searches and a quick guess as to who to notify about the delivery.
There are big complex packages that do this and everything else but cost huge amounts of money. Since this isn’t doing any complex field mapping or error checking it should be really simple to write. Really: Picture>Tesseract>Email>Done. (I know sounds simple and is simple are a world apart.) But really this almost seems like child's play to anyone who has used Tesseract already.
If adding bar code scanning isn't hard adding that would be cool, include them as text in the email body (aka BarCode1=1z234..). Also making the rescan/email choice a multiple choice rescan/email1/email2/email3 could also be a simple upgrade. But like I said simple is better.
Make it have a demo that sends up to 10 emails for free and charge $3-5 for the full version. As a business cost it is nothing, and compared to $10k for real package tracking it is less than nothing. Just set the expectation that the OCR is going to suck (because it always does) and people won't complain about it.
