ACHIEVEMENTS
AWARDS
- 2022 - Winner of Sprinter2 Programme - Smart Logistics & Buildings (HSBC)
- 2022 - Winner of HKAI Accelerator Programme (Alibaba Group)
- 2022 - Winner of Tech Raiser Programme (PwC & HK Science Park)
- 2021 - Winner of Road to GBA Challenge (PwC & HK Cyberport)
- 2020 - Gold medal at the International Exhibition of Inventions Geneva
MEDIA
- 2022 | First Edge Computing AI Chip in Hong Kong for Real-time Video Analytics
- 2022 | AI視頻分析 助監察隔離病房
- 2022 |The First Edge Computing AI Chip in Hong Kong for Real-time Video Analytics in Child Protection and Patient/Elderly Risks Monitoring during Omicron Outbreak
- 2022 | 實時視頻分析應用程式 助加強兒童保護及病患看顧
- 2022 | 首枚香港製造的邊緣計算AI芯片 助監察隔離病房
- 2022 | 實時視頻分析應用程式 可檢測人體活動及表情
- 2022 | 香港特別行政區政府資訊科技總監辦公室策劃 全港資訊科技獎勵計劃(全港首台自主研發AI人工智能計算機)
- 2021 | 人工智能时代，“芯”无止境 (智慧城市的智能心脏)
PRODUCTS
EDGE AI Chips | Lightweight DNN Framework | FiHS Runtime Libraries
Empowering AI at the Edge
We have developed a breakthrough microprocessor technology uniquely designed to accelerate AI applications on edge devices.
AI-powered features involve complex computational operations which exceed the capacity and capability of existing IC components such as CPUs and GPUs. Engineered to execute intricate computations efficiently, our chips not only can deliver the computing power required to carry out demanding DNN operations, but also can achieve real-time performance and enhance digital privacy protection and security. In addition, our lightweight DNN framework and FiHS runtime libraries can offer more powerful and efficient solution for processing massive amount of data at the edge as well as support various DNN frameworks and models. As the technology continues to advance, we believe that our novel Edge-AI technology and platform will be the key to unlock future innovations, especially in Edge Computing, AIoT, Web3, Blockchain and Metaverse.
ASIC Design
High-performance, power-efficient and cost-effective
Re-configurable Architecture
Support re-configurable CNN and FC layers
Parallel Computing
Built-in high-efficiency parallel computing micro-architecture
Customized Hardware-compatible Runtimes and Libraries
Maximize performance across processors
Unified Platform
Support DNN models trained with a variety of frameworks and tools
Visual Intelligence
Object detection, object recognition, image reconstruction, video analytics
Solutions
Data collection & pre-processing | DNN model training engine | Unified framework for model conversion | Light-weight model generation | FiHS runtime libraries and tools | Edge-based model learning and sharing | Proprietary edge AI chips for computer vision
Pattern discovery | Data mining | Machine learning | DNN | Training data preparation for AI model training | Simulated data preparation for metaverse
AI engine for image/text/speech recognition | Avatars & NPCs generation | Content and scene generation | Environment simulation
Applications
Deep Learning-based Image Reconstruction and Super-Resolution
Create a high-resolution image from a low-resolution image.
Generation of 3D Models with Deep Learning
Generate a realistic 3D model from a single 2D image.
Conversion of 2D to 3D Images with CNN
Estimate the depth map of a 2D image and generate its 3D image automatically.
Object Detection and Recognition in Real-time Video Stream
Perform real-time object detection and recognition on devices by our lightweight DNN models.
Empower the Metaverse with AI
Build the realistic metaverse with generative and emotion AI.
DNN-based Text Recognition
Extract information from unstructured documents.