計畫團隊成員
技術亮點 Technical Highlights
物聯網相關的終端裝置對於智慧視覺模組需求日漸增加。本計畫將分別開發此模組的關鍵技術,分別為:
1) 基於感測器內運算架構的低功耗高速智慧視覺感測晶片
2) 基於記憶體內運算架構且擁有高能源效率及高準確率的深度學習加速晶片
3) 仿照生物的動態視覺模型設計可追蹤物體移動的軌跡預測網路
4) 基於人腦低功耗的關鍵,設計有執行時期位元數彈性的CNN神經網路
The project will develop key technologies of the intelligent visual system as follows:
1) Low-power high-speed visual sensing chip based on processing-in-sensor architecture.
2) High accuracy and energy efficiency deep learning accelerator based on computing-in-memory architecture.
3) Spiking neural networks to predict the trajectory of moving objects based on biological dynamic vision models.
4) Convolutional neural network models with bit-width flexibility at run time based on human brain inspiration.
1) Low-power high-speed visual sensing chip based on processing-in-sensor architecture.
2) High accuracy and energy efficiency deep learning accelerator based on computing-in-memory architecture.
3) Spiking neural networks to predict the trajectory of moving objects based on biological dynamic vision models.
4) Convolutional neural network models with bit-width flexibility at run time based on human brain inspiration.