Scholarlink

The panicle count of paddy is directly associated with the yield of crop. Automation of crop yield estimation is crucial to cultivate efficient breeding techniques to fulfill the increasing population demands and adapt to climate change. Panicle count is vital in modern agriculture as a reference for precision management and plant breeding. Despite many successful studies in automatic paddy head counting, more progress has yet to be made in estimating panicle count using UAV (unmanned aerial vehicle) captured imagery, especially when the crop reaches the heading stage. Counting panicles is challenging due to many factors, such as significant variations in shape, size, posture, high density, occlusion, the effect of UAV/drone motion, etc. In this work, we implement existing state-of-the-art maize head (tassel) based counting models as benchmarks. The models, namely, TasselNet, TasselNet V2, TasselNet V2+ and TasselNet V3, were trained on 120 images and tested on 20 images. It is found that TasselNet V2+ achieves the lowest MAE (mean average error), MSE (mean square error), RMAE (root mean average error), RMSE (root mean square error) values and the highest \(R^2\) (coefficient of determination) value of 0.8547, indicating its superior performance compared to the other models. In conclusion, the trained models developed in this study could be used as a reference for panicle counting models during the vegetative and heading stages of paddy. To the best of our knowledge, this is the first of its kind that considered UAV/drone captured paddy data for panicle counting in Indian context.

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

Deep Networks Based Approach for Automatic Counting Panicles on UAV Captured Paddy RGB Imagery

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • N. Tejasri,
  • Sam Mathew Betson,
  • P. Rajalakshmi,
  • M. Balram,
  • Uday B. Desai

摘要

The panicle count of paddy is directly associated with the yield of crop. Automation of crop yield estimation is crucial to cultivate efficient breeding techniques to fulfill the increasing population demands and adapt to climate change. Panicle count is vital in modern agriculture as a reference for precision management and plant breeding. Despite many successful studies in automatic paddy head counting, more progress has yet to be made in estimating panicle count using UAV (unmanned aerial vehicle) captured imagery, especially when the crop reaches the heading stage. Counting panicles is challenging due to many factors, such as significant variations in shape, size, posture, high density, occlusion, the effect of UAV/drone motion, etc. In this work, we implement existing state-of-the-art maize head (tassel) based counting models as benchmarks. The models, namely, TasselNet, TasselNet V2, TasselNet V2+ and TasselNet V3, were trained on 120 images and tested on 20 images. It is found that TasselNet V2+ achieves the lowest MAE (mean average error), MSE (mean square error), RMAE (root mean average error), RMSE (root mean square error) values and the highest \(R^2\) (coefficient of determination) value of 0.8547, indicating its superior performance compared to the other models. In conclusion, the trained models developed in this study could be used as a reference for panicle counting models during the vegetative and heading stages of paddy. To the best of our knowledge, this is the first of its kind that considered UAV/drone captured paddy data for panicle counting in Indian context.