Scholarlink

This work proposes the design and implementation of a prototype for the automatic detection and counting of Trichoderma fungus spores through artificial vision techniques. The solution integrates both a local and an IoT-based platform, providing a user-friendly environment to support agriculture by facilitating remote monitoring and data analysis. The system is developed following the Cascade model, ensuring a structured approach throughout its various stages, from requirements gathering and design to implementation and testing with a graphical and user-friendly environment to support agriculture. Further, to achieve efficient real-time detection, the prototype employs an embedded device (specifically, an NVIDIA Jetson Nano board) enabling basic deep learning operations. A USB microscope captures spore images, which are subsequently analyzed using object detection algorithms, with Tiny YOLO selected for its lower computational overhead and reliable accuracy. This choice ensures that, despite the board’s limited resources compared to conventional servers or desktop computers, the model can effectively identify Trichoderma spores with minimal latency.

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

Agricultural Prototype for Detection and Counting of Trichoderma Fungus Spores Using Machine Vision Techniques

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Esteban Noboa-Delgado,
  • Fabián Cuzme-Rodríguez,
  • Jaime Michilena-Calderón,
  • Carlos Vásquez-Ayala,
  • Alejandra Pinto-Erazo,
  • Luis Suárez-Zambrano

摘要

This work proposes the design and implementation of a prototype for the automatic detection and counting of Trichoderma fungus spores through artificial vision techniques. The solution integrates both a local and an IoT-based platform, providing a user-friendly environment to support agriculture by facilitating remote monitoring and data analysis. The system is developed following the Cascade model, ensuring a structured approach throughout its various stages, from requirements gathering and design to implementation and testing with a graphical and user-friendly environment to support agriculture. Further, to achieve efficient real-time detection, the prototype employs an embedded device (specifically, an NVIDIA Jetson Nano board) enabling basic deep learning operations. A USB microscope captures spore images, which are subsequently analyzed using object detection algorithms, with Tiny YOLO selected for its lower computational overhead and reliable accuracy. This choice ensures that, despite the board’s limited resources compared to conventional servers or desktop computers, the model can effectively identify Trichoderma spores with minimal latency.