深度协同端计算的框架与实现技术
首发时间:2019-06-21
摘要:为进一步提升移动设备的感知能力,推动其往智能化、自主化方向发展。在现有相关研究的基础上,针对国内外研究现状中存在的典型问题,开展针对协同端计算的轻量级机器学习框架相关研究。该框架可以实现传统目标识别算法如Haar +Adaboost、HOG +SVM、等目标检测与识别方法,主要是探讨手工设计的特征对于移动平台的实用性。为了进一步提高系统的性能,结合着深度学习轻量化的目标,以海思人工智能芯片为核心的嵌入式设备为载体,开展计算机视觉如目标识别和跟踪相关理论与平台构建技术的研究。
For information in English, please click here
Framework and Implementation of Deep Collaborative Edge Computing
Abstract:In order to further enhance the perception ability of mobile devices and promote their development towards intellectualization and autonomy. We solve the challenging problems based on lightweight machine learning framework for collaborative computing on mobile devices and AI chips. This framework can implement conventional object detection methods such as Haar +Adaboost, HOG +SVM. It mainly discusses the practicability of hand-crafted features for mobile platforms. In order to further improve the performance of the system, the lightweight deep learning metFramework and Implementation of Deep Collaborative Edge Computinghods are introduced to solve the computer vision tasks such as object detection and tracking. A platform is implemented on HiSilicon Artificial Intelligence chip, which will benefit both real applications and theoretical development.
Keywords: Deep collaborative edge computing;object detection;object tracking deep learning
基金:
引用
No.****
动态公开评议
共计0人参与
勘误表
深度协同端计算的框架与实现技术
评论
全部评论0/1000