Walking into MHacks 6, our team wanted to stretch our knowledge of machine vision while utilizing Capital One's redesigned API fresh off the stack. Interestingly enough, we came across a philipino twitter account that mockingly retweets others' postings with exposed debit card information under the credo: need a debit card?
Now that online banking and transactions are commonplace, we cannot afford to let one mistake compromise a user's account. This technology isn't a consumer-ready application as much as a potential service for financial institutions to help combat fraud. Fraud costs banks heavily in time, money, and personel- not to mention reputation- and that isn't even considering how hard the victim has to stomach it.
Our service simply offers banking instutions preventative measures by beating malicious hackers to the source.
Behind The Scenes
• Pointed at any base url, our Python web crawlers and scrapers will retrieve every image url from every linked page nested within the base url.
• Afterwards, image urls are sent to the C++ image processing module that will retrieve any discernable alphanumeric characters from them thanks to a synthesis between OpenCV and Tesseract OCR.
• Appropriate banking institution will be notified of account's compromised status (i.e. Capital One).