| 摘要: |
| 随着人工智能的发展，视觉感知技术为林业害虫识别防治提供了新方法和实现思路。论文提出一种基于深度感知神经网络框架实现小蠢虫的检测识别，检测系统具体采用了特征金字塔结构、形变结构和新型非极大值抑制技术进行构建，其准确率达到了96.32%。这说明了基于人工智能技术方案识别林业害虫的可行性。与传统方法相比，此方法在精准识别林业害虫的同时，有效减少不必要的资源消耗。 |
| 关键词: 林业病虫害 小蠢虫检测识别 深度神经网络 |
| DOI： |
| 分类号: |
| 基金项目:邯郸市科学技术研究与发展计划项目(1721203049-1) |
|
| Research on Scolytidae Recognition and Identification based on ResNet Network |
|
huayueshan1, Wang Jiaxin2, Rong Jieqing2, Hua Guodong2, Li Li2
|
|
1.Guangzhou Chenjing Ecological Technology Service Co., Ltd;2.School of Information and Electrical Engineering, Hebei University of Engineering
|
| Abstract: |
| With the development of artificial intelligence, vision perception technology provides new
methods and implementation ideas for the control of forest pests. In this paper, a forest pest identification
framework based on depth perception neural network is proposed to detect and identify Scolytidae . The detection
system contains feature pyramid structure, deformable convolution structure and non maximum suppression,
which the pest identification model is 96.32%. Compared with traditional methods, this method can accurately
identify forest pests and effectively reduce the consumption of unnecessary resources. |
| Key words: forest diseases and insect pests Scolytidae recognition and identification depth neural network |
