Learning Model to Estimate Cabbage Growth Stages through Drone Observation

DOI IR

Bibliographic Information

Other Title
  • ドローン観測によるキャベツ生育ステージ推定学習モデルの提案

Search this article

Abstract

In this study, we evaluated the accuracy improvement in machine learning used for estimating the growth stages of cabbage using drone images captured at different altitudes. We adopted the YOLO model for object detection, which simultaneously identifies the position and category of objects. We constructed the learning model using images obtained at an altitude of 10 meters and the mAP@0.5 for the validation data at this altitude was found to be 0.94. However, the accuracy for validation data at an altitude of 100 meters was significantly lower at 0.156. Moreover, the model trained with images collected at an altitude of 100 meters exhibited the lowest accuracy, especially for high-resolution images. We found that high detection accuracy was achieved for both high-resolution and low-resolution images by using mixed-resolution training data.

Journal

Keywords

Details 詳細情報について

Report a problem

Back to top