Real-time

• blazing fast image recognition (up to 15 FPS) thanks to optimized SIMD computations
• optimized conversion from common camera image formats
• small image signature footprint (around 6kB per signature)

Universal

• on-device image recognition (limited by the client cache size)
• server side image recognition (scalable to million of images)
• on-device QR code / Data Matrix and EAN / UPC-A decoding

Robust

• invariant to occlusion and background clutter
• invariant to viewpoint changes
• invariant to motion blur

Lightweight

• small footprint (library ~1MB)
• standalone (no external dependency)
• 100% home made (no OpenCV, no ZXing, no ZBar)

Cross-platform

• core library written in C
• bindings for iOS (Obj-C) and Android (Java / JNI)
• plugins for RubyMotion and PhoneGap

Safe

• thread safe by design
• transactional sync
• auto-recovery in case of database corruption

Easy to use

• fully packaged sync method (no complex network code required)
• seamless scans logging in the background
• high level wrappers