SoC Design of Hand Gesture Recognition with Deep Learning

• Main Authors: Wu, Ya-Chiu, 吳雅就
• Other Authors: Tsai, Chun-Jen
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/zb7327

碩士 === 國立交通大學 === 資訊科學與工程研究所 === 106 === This thesis targets on SoC design of hand gesture recognition based on deep learning. By integrating a CNN accelerator into Xilinx Zynq-7000 SoC, we implement a full hand gesture recognition system onto SoC platform with ARM Cortex-A9 CPU. In software side, we also provide a full solution include input image pre-processing about background removal and a pure C model of CNN computation for benchmarking the accelerator. We analyze the problems and solutions when design a CNN accelerator by collecting related works about the CNN accelerator and providing the benchmark of our work for future studies.