摘要
Jetson Nano为硬件基础核心,设计并实现了目标信息检测系统。整个系统由摄像头、Jetson Nano、外围电路设备三大部分构成。通过高性能深度学习推理优化器—TensorRT技术的应用加速了神经网络的推理,实现了既能快速识别照片,又能通过摄像头实时检测目标的效果。通过实际环境的测试表明:该系统能实时完成检测视频画面的目标信息。此外,该系统具有核心处理单元体积小,能快速在多个无人系统场景部署应用的优势,且性能满足各项技术指标要求。
Using Jetson Nano as the core of the hardware, a target information detection system is designed and implemented.The whole system consists of three parts which are camera, Jetson Nano, and peripheral circuit equipment.The system accelerates neural network inference through the application of high-performance deep learning inference optimizer, TensorRT technology, and achieves the effect of not only quickly recognizing photos, but also real-time detection of targets through the camera.Through the test of the actual environment, the system can complete the target information detection of the video screen in real time.In addition, the system has the advantages of a small core processing unit, which can quickly deploy applications in multiple unmanned system scenarios, and its performance meets various technical indicators.
作者
龙诗科
蒋奇航
包友南
王建琦
LONG Shike;JIANG Qihang;BAO Younan;WANG Jianqi(School of Aeronautics and Astronautics,Guilin University of Aerospace Technology,Guilin 541004,China;Key Laboratory of Unmanned Aerial Vehicle Telemetry of Guangxi Province,Guilin 541004,China)
出处
《传感器与微系统》
CSCD
北大核心
2022年第9期99-101,108,共4页
Transducer and Microsystem Technologies
基金
国家自然科学基金资助项目(61966010)
广西小型无人机系统及应用工程研究中心资助项目
广西高校中青年教师科研基础能力提升项目(2021KY0793,2021KY0801)。
