Boxing text recognition into a cross-platform SDK
summary
Anyline becomes a cross-platform C++ SDK for live OCR on iOS, Android and Windows Phone by moving machine‑learning and vision code into a single shared C++11/14 core; that change yields one codebase, fewer platform-specific bugs, faster iteration and easier ports (a Windows Phone SDK ships in weeks/months). Implementation uses C++ core with Java wrappers, Ninja build tooling, OpenCV for vision, OpenSSL for license/networking and protobuf/JSON for data. Mobile customers include utilities, UN ID scanning and voucher/code scanning, and the SDK runs in real time on phones.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 0:44Text recognition SDK goals
- 3:08Use cases and customers
- 3:30Technical architecture overview
- 5:00Benefits of moving to C++
- 7:37Teamwork and shared codebase
- 8:45Cross-platform SDK development
- 9:30Challenges with C++
- 10:14Tools and libraries used
- 11:21Closing and contact
related talks
04mMobile text recognition for Industrial AR
37mSeeing is Believing: Mobile Vision API Deep Dive
20mImplementing Vision AI in Enterprise Android Apps
30mMobile applications that can see
48mEmpowering the frontline: Develop high-performance AI applications seamlessly for Zebra devices
40mDeveloping a Cross-Platform C++ Framework for Mobile Platforms
22mOn-Device Speech Recognition for Mobile Devices and Custom Hardware
06mProviding Better Feedback in Real-time Object Detection Apps