Android GPU Technology

Android GPU Technology

  • Android GPU Technology
    ART/170CP
    20131223 - 20150622
    14650

    Dr Jay Yeaun-Jau Liou
    1. A highly efficient video/image processing middleware on an embedded CPU-GPU platform a) A high level video/image processing middleware framework for development of media application; and b) An OpenCL—compatible highly efficient video/image processing library that is reusable and scalable to different Android OS versions and different hardware platforms. 2. Reference design (both hardware and software) of the above mentioned middleware and demo applications such as object tracking and multi-window; and 3. One patent application in one of the below three areas: 1) Embedded Parallel sequential distributed computation methodology; 2) Object tracking algorithms. 3) Usability in the embedded multi-window environment
    Mr Xiao Hao Lin Mr Yee Lim Chan Dr Yan Huo Ms Lu Wang Ms Yee Wai Mandy Yim Mr Hanqiang Huang Mr Pan Lit Wong Mr Chi Chung Leung Ms Hua Yang Mr Fu Chun Li Mr Kwok Hung Leung Mr Yat Cheung Ngai Ms Yin Yee Chan Mr Sheng Zhou Mr Ka Yuk Lee Mr Chun Chung Lo Mr Tony Wu Mr RiHong Chen Mr Denny Lai Mr Teddy S. H. CHU Mr Wong Tin Lung Mr LaiFa Fang Dr SaiFan, Dominic Chan Mr QiangQi Zou Mr William Wai Lun, Lee Mr Ken Ka Fai Suk Mr Edmund Ching Lun Lo Mr Jeffrey Ding Lap Cheung Mr Jackie Chi Hing Chau
    Aircraft Medical Ltd [Sponsor] Clever Motion Technology Ltd (License Income) [Sponsor] Hong Kong Linux Industry Association KoolSee Medianet Corp Limited KoolSee Medianet Corp Limited (License Income) [Sponsor]

    Riding on the fast growing trend of low-cost, high-performance, multi-core Android SoC, this project aims to develop a video/image processing middleware technology platform for embedded systems(comparable to DirectX for PC) utilizing the heterogeneous CPU-GPU architecture of multi-core SoC. This technology will be hardware and Android version independent. It will provide unified APIs to application developers with rich video/image, camera and machine learning libraries extracted from academic research and ASTRI patent portfolio. It enables fast and efficient commercialization of academic/university research. Demo applications such as object tracking and multi-window will be developed in this project using the above mentioned middleware technology platform for proof-of-technology, innovative MMI applications, and early commercialization engagements.