基于改进YOLO V5的无人机航拍图像小目标检测方法 |
| |
作者姓名: | 程向红 曹毅 胡彦钟 张文卓 钱荣辉 |
| |
作者单位: | 东南大学 仪器科学与工程学院
微惯性仪表与先进导航技术教育部重点实验室 |
| |
基金项目: | 仪器科学与工程学院2021年国家级/省级SRTP项目(5222002102B) |
| |
摘 要: | 本文针对无人机航拍目标检测技术中目标聚集、目标较小及实时性差等问题,将YOLO V5的主干架构进行改进,简化Neck网络,提出了一种提高检测速度又能准确识别的无人机航拍图像检测技术方案。经过仿真实验测试,改进后的YOLO V5网络在保持识别精度的同时,检测速率提升了31%,满足无人机在航拍作业时对于准确性与实时性的要求。
|
关 键 词: | 深度学习 无人机航拍 目标检测 |
Target Detection Method of UAV Aerial Image Based on Improved YOLO V5 |
| |
Authors: | CHENG Xianghong CAO YI HU Yanzhong ZHANG Wenzhuo QIAN Ronghui |
| |
Institution: | School of Instrument Science and Engineering, Southeast University;
Key Laboratory of Micro-Inertial Instrument and Advanced Navigation Technology, Ministry of Education |
| |
Abstract: | Aiming at the problems such as target aggregation, small target and poor real-time performance of UAV aerial target detection technology, this paper improves the backbone architecture of YOLO V5, simplifies the Neck network, and proposes a UAV aerial image detection technology scheme that can improve detection speed and accurately identify at the same time. Through the simulation test, the improved YOLO V5 network has improved the detection rate by 31% while maintaining the recognition accuracy, which meets the requirements of accuracy and real-time performance of UAVs in aerial photography. |
| |
Keywords: | deep learning UAV aerial photography target detection |
|
| 点击此处可从《》浏览原始摘要信息 |
| 点击此处可从《》下载免费的PDF全文 |