期刊文献+
共找到3篇文章
< 1 >
每页显示 20 50 100
资源受限MCU的轻量化部署策略和实现
1
作者 吴薇 阮星 +3 位作者 蔡闯华 刘长勇 刘彦秀 王宜怀 《吉林大学学报(信息科学版)》 CAS 2023年第6期1063-1071,共9页
为实现低资源嵌入式设备的图像分类识别,针对能实现简单图像识别任务、对图像识别准确率要求不高,且要求低成本的场景,将卷积神经网络(CNN:Convolutional NeuralNetwork)部署到资源受限的微控制器单元(MCU:Microcontroller Units)上。... 为实现低资源嵌入式设备的图像分类识别,针对能实现简单图像识别任务、对图像识别准确率要求不高,且要求低成本的场景,将卷积神经网络(CNN:Convolutional NeuralNetwork)部署到资源受限的微控制器单元(MCU:Microcontroller Units)上。首先提出一种在资源受限MCU上的轻量化部署策略:为降低模型的参数量,提出一种轻量化的神经网络算法;为保证模型大小能适应有限的随机存取存储器(RAM:Random Access Memory),提出了一种基于闪存(FLASH:Flash Memory)扇区的替存储算法。其次,在资源受限的嵌入式设备上部署该策略。针对采集图像的质量和采集速度不匹配问题,设计了摄像头外围电路;对采集图像进行基于高斯分布的自适应阈值二值化处理并对图像样本完整性进行校验。实验结果表明,该系统取得大约80%~89%的识别准确率。虽然该准确率低于训练精度10%左右,但在上述对精度要求不高的实际场景中可以较好地应用。 展开更多
关键词 嵌入式系统 资源受限MCU 图像识别 深度学习
下载PDF
实时操作系统mbedOS启动流程剖析 被引量:9
2
作者 刘长勇 王宜怀 +1 位作者 蔡闯华 蒋建武 《计算机工程与应用》 CSCD 北大核心 2020年第20期46-51,共6页
mbedOS是ARM公司于2014年开始推出的一款面向智能终端与物联网节点的实时操作系统,它提供了线程管理与调度、内存管理、时钟管理、队列管理等基本功能要素。在深入分析mbedOS的功能要素和函数调用关系的基础上,简要给出了KL36芯片的启... mbedOS是ARM公司于2014年开始推出的一款面向智能终端与物联网节点的实时操作系统,它提供了线程管理与调度、内存管理、时钟管理、队列管理等基本功能要素。在深入分析mbedOS的功能要素和函数调用关系的基础上,简要给出了KL36芯片的启动流程,重点剖析了mbedOS堆栈的设置、中断向量表的重定向、内核的初始化、主线程的创建以及内核的启动,最后对mbedOS启动过程中存储使用情况进行了分析,为mbedOS在不同微控制器上的移植提供了技术基础,也可为分析其他RTOS的启动提供参考。 展开更多
关键词 mbedOS 启动流程 线程 KL36
下载PDF
Spatial and temporal dynamics of phytoplankton and bacterioplankton biomass in Sanya Bay, northern South China Sea 被引量:11
3
作者 ZHOU Weihua LI Yao +5 位作者 cai chuanghua HUANG Liangmin WANG Hankui XU Jirong DONG Junde ZHANG Si 《Journal of Environmental Sciences》 SCIE EI CAS CSCD 2009年第5期595-603,共9页
The composition of phytoplankton and the dynamics of phytoplankton and bacterioplankton biomass (PB and BB, respectively) of Sanya Bay, South China Sea, were determined. A total of 168 species (67 genera) phytopla... The composition of phytoplankton and the dynamics of phytoplankton and bacterioplankton biomass (PB and BB, respectively) of Sanya Bay, South China Sea, were determined. A total of 168 species (67 genera) phytoplankton were identified, including BaciUariophyta (diatom, 128 species), Pyrrophyta (35 species), Cyanophyta (3 species), and Chrysophyta (2 species). Annual average zbundance of phytoplankton was 1.2 × 107 cells/m^3, with the highest abundance in autumn, and the lowest in summer. Annual average diversity index (H') and evenness (J) values were 3.96 and 0.70, respectively. Average chlorophyll-a was 2.5 mg/m^3, and the average PB was 124 mg C/m^3, with the highest value in autumn. Surface PB was higher than the bottom, except for summer. Annual mean bacterioplankton abundance and BB were 6.9 × 10^1l cells/m^3 and 13.8 mg C/m^3, respectively. The highest BB was found in summer, followed by winter, spring, and autumn. Surface BB was higher than bottom all year round. The spatial distribution patterns of PB and BB were very similar with the highest biomass in the estuary, and decreased seaward, primarily due to the terrestrial input from the Sanya River and influx of oceanic water. The main factor influencing PB and BB was dissolved inorganic nitrogen (DIN). Other factors such as temperature, which is above 22℃ throughout the year, had a negligible impact. The correlation between BB and PB was significant (P 〈 0.01). The annual average ratio of BB/PB was 0.12 (0.06-0.15). Phytoplankton primary production was one of the most important factors in controlling the distribution of bacterioplankton. 展开更多
关键词 BIOMASS PHYTOPLANKTON BACTERIOPLANKTON Sanya Bay northern South China Sea
下载PDF
上一页 1 下一页 到第
使用帮助 返回顶部