Scholarlink

<p>To address the challenges of low efficiency and high cost in monitoring invasive plants in ecologically sensitive areas such as nature reserves, wetlands, and grasslands, this study has designed and developed an intelligent mobile offline recognition solution integrating the Adaptive Hybrid Feature Correction-Guard Network (AHFC-Guard Network). Through algorithm innovation and engineering optimization, this solution has built a mobile terminal recognition system suitable for complex environments, capable of identifying 38 major invasive plant species with an invasion level rated as grade I within China, providing a portable and efficient solution for ecological security protection. At the algorithm level, a novel Adaptive Hybrid Feature Correction Module (AHFCM) integrating the Hybrid Feature Correction Module (HFCM) and the Adaptive Channel Statistical Path (ACSP) is proposed. By cascading three AHFCMs to form the AHFC-Guard Network, it effectively suppresses complex background noise and enhances the stable discrimination between species. The experimental results show that the model has a parameter size of only 0.941&#xa0;M and achieves an average recognition accuracy of 97.75% for 38 types of invasive plants. At the application function level, an offline mobile application for fieldwork has been developed. The installation package size is only 67.6&#xa0;MB and it supports offline display of the Top-3 species recognition results. It embeds a database of 38 types of invasive plants' hazard characteristics, supporting graphic and text comparison queries for morphological features, ecological impacts, and control plans. This research provides full-chain technical support from species identification, hazard assessment to control decision-making for ecological protection departments. Through a high-precision and low-cost mobile monitoring solution, it significantly enhances the biosecurity protection efficiency in ecologically sensitive areas.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A mobile identification system for invasive plants based on adaptive hybrid feature correction-guard network model, supporting fully offline decision-making

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年12月5日

查看全文
  • Yan Ren,
  • Guoxin Li,
  • Zhi Li,
  • Wenli An,
  • Jiarong Yang,
  • Haiming Ni

摘要

To address the challenges of low efficiency and high cost in monitoring invasive plants in ecologically sensitive areas such as nature reserves, wetlands, and grasslands, this study has designed and developed an intelligent mobile offline recognition solution integrating the Adaptive Hybrid Feature Correction-Guard Network (AHFC-Guard Network). Through algorithm innovation and engineering optimization, this solution has built a mobile terminal recognition system suitable for complex environments, capable of identifying 38 major invasive plant species with an invasion level rated as grade I within China, providing a portable and efficient solution for ecological security protection. At the algorithm level, a novel Adaptive Hybrid Feature Correction Module (AHFCM) integrating the Hybrid Feature Correction Module (HFCM) and the Adaptive Channel Statistical Path (ACSP) is proposed. By cascading three AHFCMs to form the AHFC-Guard Network, it effectively suppresses complex background noise and enhances the stable discrimination between species. The experimental results show that the model has a parameter size of only 0.941 M and achieves an average recognition accuracy of 97.75% for 38 types of invasive plants. At the application function level, an offline mobile application for fieldwork has been developed. The installation package size is only 67.6 MB and it supports offline display of the Top-3 species recognition results. It embeds a database of 38 types of invasive plants' hazard characteristics, supporting graphic and text comparison queries for morphological features, ecological impacts, and control plans. This research provides full-chain technical support from species identification, hazard assessment to control decision-making for ecological protection departments. Through a high-precision and low-cost mobile monitoring solution, it significantly enhances the biosecurity protection efficiency in ecologically sensitive areas.