Unique in its kind AI-powered automated data capture software.
ChallengeAdd camera support to Android application and fine-tune data gathering method via the camera to collect data for OCR
SolutionOCR algorithm to scan hidden code with infrared
Technologies and toolsBackend: Python, OpenCV, Tesseract OCR; Android: Android SDK, Java, USB serial protocol for Android, USBLIB+UCV Camera Source Code, and other
The Client is a startup that works in the niche of security printing and delivers proprietary technology. The business is aimed at brand owners who need extra security measures to confront counterfeiters in their market. The Client’s technology enables printing of security marks on whatever surface. The watermarks become visible and can be scanned in infrared only. The Client was in search of a reliable provider of OCR software development services to tailor a solution for Android phones to scan and recognize printed marks.
Challenge: add camera support to Android application and fine-tune data gathering method via the camera to collect data for OCR
The Client needed a custom OCR solution to enable infrared scanning and recognition capabilities. The algorithm had to be deployed on mobile devices powered by Android to allow instant online verification of security marks.
The OCR algorithm had to encompass the following stages: capture and recognize characters, send a code to the server, and conduct validation. The ESSID Solutions’ team was challenged to develop a PoC, that also included Android app for OCR, in line with all the requirements.
Solution: OCR algorithm to scan hidden code with infrared
Our team had to augment the existing mobile app through building custom OCR software for Android phones. A device was linked by USB to PCB (printed circuit board) embedded into the camera for infrared scanning.
The mobile app augmented by the OCR algorithm was expected to perform the following tasks:
- Live stream from the camera
- Sent image to the ESSID Solutions’ backend for processing (image enhancement/ROI detection/autorotation/OCR)
The ESSID Solutions’ team used open-source software, such as OpenCV library and Tesseract OCR engine, and applied innovative approaches to customize the solution according to the specific needs. The data gathering was conducted via an IR camera.
ESSID Solutions’ engineers faced major challenges during the phase of data gathering:
- There was no solution available to support the camera for Android device
- It was required to build the backend solution to work coherently with the Android software and the Client’s database
The algorithm enabled the Image Auto Capture mode at the configurable distance and at various angles. The application uses API by ESSID Solutions’ for image enhancement.
The original image is a set of 15 figures that can be captured at the wrong angle. A captured image then goes to the ESSID Solutions’ backend for autorotation. The proprietary OCR algorithm scans the image, receives a code, and sends the code to the Client’s API for good/bad evaluation.
Result: high-quality backend to collect data of needed quality for OCR
The Client had specific marks for scanning in the infrared and needed OCR application development services to solve this challenging task. Our team had to put efforts into adjusting cameras for specific lighting. Also, the engineering team has solved the task of linked the camera with the OCR algorithm to enable the collection of quality input data and image enhancement capabilities.
Our team has successfully built and delivered automated data capture software to augment the existing solution with AI. The delivered system is the ESSID Solutions’ proprietary custom solution, the only working solution compatible with Android software.