Vase-Life Monitoring System for Cut Flowers Using Deep Learning and Multiple Cameras.

Abstract

Here, we developed a vase-life monitoring system (VMS) to automatically and accurately assess the post-harvest quality and vase life (VL) of cut roses. The VMS integrates camera imaging with the YOLOv8 (You Only Look Once version 8) deep learning algorithm to continuously monitor major physiological parameters including flower opening, fresh weight, water uptake, and gray mold disease incidence. Our results showed that the VMS can automatically measure the main physiological factors of cut roses by obtaining precise and consistent data. The values measured for physiology and disease by the VMS closely correlated with those measured by observation (OBS). Additionally, YOLOv8 achieved a high performance in the model by obtaining an object detection accuracy of 90%. Additionally, the mAP0.5 supported the high accuracy of the model in evaluating the VL of cut roses. Regression analysis revealed a strong correlation between the VL, VMS, and OBS. The VMS incorporating the microscope detected physiological and disease factors in the early stages of development. These results show that the plant monitoring system incorporating a microscope is highly effective for evaluating the post-harvest quality of cut roses. The early detection method using the VMS could also be applied to the flower breeding process, which requires rapid measurements of important characteristics of flower species, such as VL and disease resistance, to develop superior cultivars.

Plant phenotyping relevance

カメラ画像とYOLOv8を統合したシステムを開発・検証し、切り花の開花、鮮重、吸水、灰色かび病、花瓶寿命を自動評価しているため、植物表現型の取得手法が中心です。

Code and data availability

The article describes a vase-life monitoring system with YOLOv8 models and a 50,000-image dataset, but no public deposit of the dataset, images, trained models, or analysis code is provided. The only supplement (Supplementary Table S1) contains a model accuracy comparison, not data, code, or images. No authors' public,

No evidence-backed public reproduction asset is currently recorded.