Skip to content

Welcome to AI System Lab

本實驗室專注於人工智慧與智慧運算的前瞻研究，涵蓋3D人體建模、電腦視覺、動態視覺感測、時序資料分析、模型壓縮中介軟體，以及AI專用硬體架構等六大領域。我們強調端到端的系統開發，從演算法設計到硬體實現，致力於軟硬體協同設計，打造硬體友善且高能源效率（energy efficient）的AI解決方案。研究團隊積極推動AI模型與加速器的深度整合，實現低功耗、高效率的智慧應用，並將創新技術落實於實際場域，促進AI技術的產業化與普及。

研究領域

  • 3D Human Modeling

    • Space Attention
    • Un-calibrated Camera
    • Real-world Mesh Reconstruction
    • 體智能系統開發
    • Model Compression

  • Computer Vision (CV) Applications

    • Multi-Task Learning
    • Gaze Estimation
    • Object Tracking
    • Monocular Depth Estimation
    • Object Detection
    • Adversarial Attack for Object Detection
    • Defect Detection

  • Dynamic Vision Sensor (DVS) Applications

    • Depth Estimation
    • Segmentation
    • Event-based Human Modeling

  • Time Series Applications

    • Audio Classification
    • Keyword Spotting
    • Gravitation Wave

  • Model Compression and Middleware

    • CNN Pruning
    • CNN Quantization
    • Transformer Pruning
    • Transformer Quantization
    • Transformer Compression
    • AI Compiler
    • Simulator and Emulator
    • Virtual Platform and Tools

  • AI Hardware Architecture and Acceleration

    • CNN Processor
    • RISC-V CPU
    • Transformer Processor
    • Multi-modal Processor