Web App with Accurate OCR: Fast Scanning and Efficient Management of Fiscal Receipts
JavaScript is disabled in your browser.
Please enable JavaScript settings in your browser to allow this website work correctly.
Customer
Epson
Problem
Epson was looking to augment its new cutting-edge product with a powerful desktop software to be supplied along with the device — a next-gen photo scanning solution. The software was to enable fast processing of large batches of analog photoprints to store them as a digital gallery under a user’s account — either local or Internet-based.
With proven real-life software engineering experience under its belt demonstrated during past collaborations with Epson, Softeq was engaged to take on the solution implementation.
Softeq was to implement a desktop app supplied with an Epson scanner to enable fast analogue photo processing and more convenient image management in line with the latest UX trends.
Solution
The team created a desktop app installable upon the initial device connection to a computer. The scanner is equipped with a 30-photo auto feeder to scan photos, both the front and back, to a designated hard drive, while also offering tools to support easy uploading to social media (e.g. Facebook, Dropbox® and Google DriveTM). The scanning software features a unique naming system enabling to organize photo collections with recognizable file and folder names, searchable metadata, and the image capture date.
Key Features:
- Automatically installs itself, the device driver and other designated Epson software
- Scans photos (by batches up to 30 items per load) into custom named folders with custom filenames
- Allows for automatic (right upon scanning) and manual image enhancement, color restoration and red-eye reduction
- Supports image editing and tagging
- Enables Images sharing via social media
- Allows uploading images to Google Drive and Dropbox
Unique Red Eye Remover
To deliver the required image processing functionality the team developed three algorithms:
- Photo enhancement algorithm
- Color restoration algorithm
- Red eye reduction algorithm
The latter appears to have no analogs on the market in terms of full-cycle automation — from face recognition through eyes and pupil detection to red pupil retouch. Recognizing the glowing red pupil and creating a mask for pupil retouch exactly matching its shape was the most challenging task, which the team resolved successfully. The resulting solution has 70% accuracy for 300 dpi photo scans disregarding the photo orientation, head position and eye shape.
Security
The security is provided through the authentication mechanisms enabled by the respective social media APIs.
Methodology
The team employed a conventional Waterfall-based software development process due to well-defined requirements and scope availability.
Results
Happy about the delivery, Epson has successfully launched the sales of the brand new device.
Unlike scanning photos on a flatbed scanner, which can take significant amount of time, or sending photos out to a costly scanning service, the complete scanning system offers a fast, easy and affordable solution for preserving and sharing a family’s legacy.
