基于深度学习的无人机森林虫害图像小目标检测系统V1.0

随着计算机分析技术的发展，在对森林虫害检测中，常利用无人机采集林场样地的正射图像，结合计算机技术进行分析。为保证检测的时效性和准确性，需要一个能真正实现小目标检测的无人机检测系统。
基于深度学习的森林虫害无人机小目标检测系统以受红脂大小蠹危害的油松作为实验目标，无需航片拼接、正射校正等预处理，通过在林区无人机遥控现场架设移动图形工作站，完成对受害油松的小目标检测。该系统包括航拍无人机、Android无人机遥控器和移动图形工作站三部分。其中部署在移动图形工作站的检测模型，是通过在服务器训练精简的DSFD目标检测框架所得。
基于深度学习的森林虫害无人机小目标检测系统利用无人机航拍，在林区无人机遥控现场可快速、有效检测出受害油松，实现对森林虫害的小目标检测。

With the advancement of computer analysis technologies, unmanned aerial vehicles (UAVs) are often used to collect orthorectified images of forest farm sample plots for forest pest detection, combined with computer technologies for analysis. To ensure the timeliness and accuracy of detection, a UAV-based detection system that can truly realize small-object detection is required.

The deep learning-based UAV small-object detection system for forest pest detection takes Chinese red pines (Pinus tabuliformis) damaged by red turpentine beetles (Dendroctonus valens) as experimental subjects. Without the need for preprocessing steps such as aerial image mosaicking and orthorectification, it completes small-object detection of damaged Chinese red pines by setting up a mobile graphics workstation at the UAV remote control site in the forest area. The system consists of three parts: an aerial photography UAV, an Android-based UAV remote controller, and a mobile graphics workstation. The detection model deployed on the mobile graphics workstation is obtained by training a streamlined DSFD object detection framework on a server.

This deep learning-based UAV small-object detection system for forest pest detection uses UAV aerial photography to quickly and effectively detect damaged Chinese red pines at the UAV remote control site in the forest area, realizing small-object detection of forest pests.

搜集汇总

背景与挑战

背景概述

该数据集是一个基于深度学习的无人机森林虫害图像小目标检测系统软件，主要用于检测受红脂大小蠹危害的油松，无需航片拼接等预处理，可在林区现场快速实现小目标检测。它属于'人工林重大灾害防控关键技术研究'项目，数据量为238.02 MB，发布于2023年，采用协议共享方式提供。