摘要
提出了一种使用基于RT-Thread实时操作系统和STM32L431微控制器的嵌入式识别系统。使用VGA摄像头实时的采集被识别物体的图像信息,对采集后的图像信息进行图像预处理;然后使用MobileNetV3模型进行参数训练;最后,在嵌入式终端实现了对图像的推理识别。实验表明,对MobileNetV3模型的验证和在终端进行实物测试的结果性能指标高、响应时间快。
This article proposes an embedded recognition system based on RT Thread real-time operating system and STM32L431 microcontroller.A VGA camera is used to collect real-time image information of the identified object,and image preprocessing is performed on the collected image information;Then,the MobileNetV3 model is used for parameter training,and finally,inference recognition of images is implemented on an embedded terminal.The experiment shows that the validation of the MobileNetV3 model and physical testing on the terminal have the result in high performance indicators and fast response time.
作者
张烁
张荣
Zhang Shuo;Zhang Rong(Department of Computer and Information Engineering,Shanxi Youth Vocational College,Taiyuan Shanxi 030032,China;Computer Teaching Department,Shanxi Medical University,Taiyuan Shanxi 030001,China)
出处
《山西电子技术》
2024年第5期86-88,98,共4页
Shanxi Electronic Technology
基金
国家自然科学基金(81973154)
山西省自然科学基金(201901D111204)。
关键词
嵌入式
图像识别
实时系统
深度学习
embedded
image recognition
real-time system
deep learning
