Testing videos about AR. All versions are only based on the OpenCV framework without commerial SDKs.
Left-up : The mobile-ar-project transplanted from takmin's desktop-ar-project.(detection SLOW/with Accumulated errors)
detection 80ms/tracking 40ms, open sourced below.
Middle-up : The detection and tracking sample for current company.(FAST-detection/Tracking accurate-pose) future work: robust pose with fast-moving/robust detection&tracking with small-scale)
Currently with only single-thread-implemention, the speed is detection 10ms/tracking 10ms on iPhone5C, which cost 1ms on iMac.
Right-up : multi-detection & tracking test.
Left-down : tracking sample based on Linear Predictor Learning: Very Fast tracking speed,robust on blur. (pose&light-change-invariant need to be improved)
The detection/tracking/GLES render(also with skeleton animation) are developed with OpenCV as dependency only, which means the ar code can be tranplanted to any platform.
mobile-AR upon takmin's project(source opend).
The detection and tracking sample in development.
Multiple detection and tracking sample in development.
The Linear Predictor tracking implemention sample.