All Issue

2026 Vol.35, Issue 2 Preview Page

Original Articles

Time-series Deep Learning Model for Predicting Marigold Flower Blooming Progress

매리골드 꽃의 개화 예측을 위한 시계열 기반 딥러닝 모델

Nayoung Kwak1
,
Bo Hyun Sung1
,
Wook Oh23
,
Young-Yeol Cho23*
곽 나영1
,
성 보현1
,
오 욱23
,
조 영열23*
1Graduate Student, Department of Horticultural Science, Jeju National University, Jeju 63243, Korea
2Professor, Department of Horticultural Science, Jeju National University, Jeju 63243, Korea
3Researcher, Research Institute for Subtropical Agriculture and Animal Biotechnology, SARI, Jeju National University, Jeju 63243, Korea
1제주대학교 생명자원과학대학 원예학과 대학원생
2제주대학교 생명자원과학대학 원예학과 교수
3제주대학교 아열대농업생명과학연구소, 친환경연구소 연구원
*Corresponding Author
30 April 2026. pp. 154-160
PDF XML Citation
Abstract

This study proposes a deep learning-based approach to quantitatively model the flowering progression of French marigold (Tagetes patula L.) using image data. The experiment was conducted under a controlled plant factory environment with hydroponic cultivation, and front-facing images were taken at hourly intervals from the early flowering stage to full bloom. To capture temporal dynamics, five consecutive images per plant were grouped as one input sequence, and the model was trained to predict the flowering progress (ranging from 0 to 1) at the final time step. The proposed model adopts a hybrid CNN+RNN architecture that combines convolutional neural networks (CNN) for feature extraction and long short-term memory (LSTM) networks to reflect temporal patterns. For performance comparison, four baseline CNN-based models—BasicCNN, ResNet18, MobileNetV2, and EfficientNetB0—were evaluated under the same dataset and conditions. Experimental results demonstrate that the CNN+RNN model significantly outperformed the other models, achieving a mean squared error (MSE) of 0.0006, a mean absolute error (MAE) of 0.0196, and an R2 value of 0.993. These findings suggest that the proposed model effectively captures temporal visual changes and can serve as a foundational technique for automated flowering monitoring and phenological analysis in floriculture crops.

Keywords
CNN+RNN hybrid
deep learning
flowering prediction
growth stage estimation
plant phenotyping
Tagetes patula L.

본 연구는 프렌치매리골드(Tagetes patula L.)의 개화 과정을 영상 기반으로 정량화하고, 이를 바탕으로 개화 진행률을 예측할 수 있는 딥러닝 모델을 제안하고자 하였다. 실험은 완전제어형 식물공장에서 수경재배 조건하에 수행되었으며, 개화 초기부터 만개에 이르기까지 1시간 간격으로 정면 이미지를 촬영하여 시계열 이미지 데이터셋을 구축하였다. 각 식물체의 이미지는 시간순으로 정렬된 연속 이미지 5장을 하나의 시퀀스로 구성하고, 마지막 시점의 개화율을 예측하는 회귀 문제로 모델을 학습하였다. 제안한 모델은 CNN 기반 특징 추출기와 LSTM 기반 RNN을 결합한 CNN+RNN 하이브리드 구조로, 시간에 따른 개화 과정을 효과적으로 반영하도록 설계되었다. 비교 실험에는 BasicCNN, ResNet18, MobileNetV2, EfficientNetB0 모델을 포함하였으며, 동일한 데이터셋과 조건에서 성능을 비교하였다. 실험 결과, 제안한 CNN+RNN 모델은 평균제곱오차(MSE) 0.0006, 평균절대오차(MAE) 0.0196, 결정계수(R2) 0.993을 기록하며, 다른 모델들에 비해 데이터에 대한 매우 높은 적합도를 나타냈다. 이러한 결과는 제안된 모델이 시간에 따른 시각적 변화를 효과적으로 반영함을 시사하며, 화훼 작물의 자동 개화 모니터링 및 생육 단계 분석을 위한 기반 기술을 활용될 수 있음을 보여준다.

키워드
개화 예측
딥러닝
생장 단계 추정
식물 표현형 분석
CNN+RNN 하이브리드 모델
Tagetes patula L.
References
1

Bhave A, Schulzová V, Mrnka L, Hajšlová J (2020) Influence of harvest date and postharvest treatment on carotenoid and flavonoid composition in French marigold flowers. J Agric Food Chem 68:7880-7889. https://doi.org/10.1021/acs.jafc.0c02042

10.1021/acs.jafc.0c02042
2

Ghosal S, Zheng B, Chapman SC, Potgieter AB, Jordan DR, Wang X, Singh AK, Singh A, Hirafuji M et al (2018) A weakly supervised deep learning framework for sorghum head detection and counting. Plant Phenomics 2019:1525874. https://doi.org/10.34133/2019/1525874

10.34133/2019/152587433313521PMC7706102
3

He K, Zhang X, Ren S, Sun J (2016) Deep residual learning for image recognition. CVPR:770-778. https://doi.org/10.1109/CVPR.2016.90

10.1109/CVPR.2016.90
4

Kim MH, Song CE, Park YJ (2021) Study on the quality of marigold (Tagetes erecta L.) flower tea by its manufacture methods for expansion of utilization of marigold as tourism resources. J Korean Soc Rural Tour 24:75-89

5

Jiao G, Shentu X, Zhu X, Song W, Song Y, Yang K (2022) Utility of deep learning algorithms in initial flowering period prediction models. Agriculture 12:2161. https://doi.org/10.3390/agriculture12122161

10.3390/agriculture12122161
6

Kamilaris A, Prenafeta-Boldú FX (2018) Deep learning in agriculture: A survey. Comput Electron Agric 147:70-90. https://doi.org/10.1016/j.compag.2018.02.016

10.1016/j.compag.2018.02.016
7

Kishimoto S, Maoka T, Sumitomo K, Ohmiya A (2005) Analysis of carotenoid composition in petals of calendula (Calendula officinalis L.). Biosci Biotech Biochem 69:2122-2128. https://doi.org/10.1271/bbb.69.2122

10.1271/bbb.69.2122
8

Lee MK, Park JS, Song HJ, Chon SU (2014) Effects of polyphenol and catechin levels on antioxidant activity of several edible flower extracts. Korean J Plant Res 27:111-118. https://doi.org/10.7732/kjpr.2014.27.2.111

10.7732/kjpr.2014.27.2.111
9

Lv M, Xu W, Chen T (2019) A hybrid deep convolutional and recurrent neural network for complex activity recognition using multimodal sensors. Neurocomputing 362:33-40. https://doi.org/10.1016/j.neucom.2019.06.051

10.1016/j.neucom.2019.06.051
10

Meda A, Lamien CE, Romito M, Millogo J, Nacoulma OG (2005) Determination of the total phenolic, flavonoid and proline contents in Burkina Fasan honey, as well as their radical scavenging activity. Food Chem 91:571-577. https://doi.org/10.1016/j.foodchem.2004.10.006

10.1016/j.foodchem.2004.10.006
11

Muley BP, Khadabadi SS, Banarase NB (2009) Phytochemical constituents and pharmacological activities of Calendula officinalis Linn (Asteraceae): a review. Trop J Pharm Res 8:455-465. https://doi.org/10.4314/tjpr.v8i5.48090

10.4314/tjpr.v8i5.48090
12

Park JW, Kim WS (2017) Effect of UV-B radiation on antioxidation in petals of edible flower pansy. Flower Res J 25:24-28. https://doi.org/10.11623/frj.2017.25.1.04

10.11623/frj.2017.25.1.04
13

Sandler M, Howard A, Zhu M, Zhmoginov A, Chen L-C (2018) MobileNetV2 : Inverted residuals and linear bottlenecks. CVPR:4510-4520. https://doi.org/10.1109/CVPR.2018.00474

10.1109/CVPR.2018.00474
14

Siriamornpun S, Kaisoon O, Meeso N (2012) Changes in colour, antioxidant activities and carotenoids (lycopene, β-carotene, lutein) of marigold flower (Tagetes erecta L.) resulting from different drying processes. J Funct Foods 4:757-766. https://doi.org/10.1016/j.jff.2012.05.002

10.1016/j.jff.2012.05.002
15

Tan M, Le QV (2019) EfficientNet : Rethinking model scaling for convolutional neural networks. ICML:6105-6114. https://doi.org/10.48550/arXiv.1905.11946

10.48550/arXiv.1905.11946
16

Thomas B, Vince-Prue D (1997) Photoperiodism in plants. 2nd ed. Academic Press.

17

Yoo IS, Park JM, Kim AJ (2021) Evaluation of the physicochemical activities of frequently consumed edible flower teas in Korea. Asian J Beauty Cosmetol 19:289-301. https://doi.org/10.20402/ajbc.2021.0173

10.20402/ajbc.2021.0173
18

Ghosal S, Zheng B, Chapman SC, Potgieter AB, Jordan DR, Wang X (2018) A weakly supervised deep learning framework for sorghum head detection and counting. Plant Methods 14:1-10. https://doi.org/10.1186/s13007-018-0300-5

10.1186/s13007-018-0300-5
19

Ma Q, Liu Z, Zheng Z, Huang Z, Zhu S, Yu Z, Kwok JT (2024) A survey on time-series pre-trained models. IEEE Trans Knowl Data Eng 36:7536-7555. https://doi.org/10.1109/TKDE.2024.3475809

10.1109/TKDE.2024.3475809
20

Mann HM, Iosifidis A, Jepsen JU, Welker JM, Loonen MJ, & Høye TT (2022) Automatic flower detection and phenology monitoring using time‐lapse cameras and deep learning. Remote Sens Ecol Conserv 8:765-777. https://doi.org/10.1002/rse2.275

10.1002/rse2.275
21

Lee SH, Kim JY, Lee SW, et al. (2014) Quality characteristics and physiological activity of edible flowers. Korean J Food Preserv 21:912-920. https://doi.org/10.11002/kjfp.2014.21.6.912

10.11002/kjfp.2014.21.6.912
22

Liu L, Yin X, Zhang X, Wang Q, Shi Y, Wang Y (2021) Plant growth stage classification using deep learning and vision transformers. Comput Electron Agric 190:106500. https://doi.org/10.1016/j.compag.2021.106500

10.1016/j.compag.2021.106500
23

Sankaran S, Khot LR, Espinoza CZ, Jarolmasjed S, Sathuvalli VR, Vandemark GJ, Miklas PN, Carter AH, Pumphrey MO, Knowles NR (2015) Low-altitude, high-resolution aerial imaging systems for row and field crop phenotyping: a review. Eur J Agron 70:112-123. https://doi.org/10.1016/j.eja.2015.07.004

10.1016/j.eja.2015.07.004
24

Singh A, Ganapathysubramanian B, Sarkar S, Singh AK (2020) Deep learning for plant stress phenotyping: Trends and future perspectives. Trends Plant Sci 25:883-898. https://doi.org/10.1016/j.tplants.2018.07.004

10.1016/j.tplants.2018.07.004
25

Yu X, Pan W, Liu G, Wang Q, Jiang Y, Han Y (2022) A CNN-LSTM model for plant growth stage classification in precision agriculture. Sensors 22:917. https://doi.org/10.3390/s22030917

10.3390/s2203091735161663PMC8839390
Information
  • Publisher :The Korean Society for Bio-Environment Control
  • Publisher(Ko) :(사)한국생물환경조절학회
  • Journal Title :Journal of Bio-Environment Control
  • Journal Title(Ko) :생물환경조절학회지
  • Volume : 35
  • No :2
  • Pages :154-160
  • Received Date : 2025-12-10
  • Revised Date : 2026-04-28
  • Accepted Date : 2026-04-30
  • DOI :https://doi.org/10.12791/KSBEC.2026.35.2.154
Journal Informaiton Journal of Bio-Environment Control Journal of Bio-Environment Control
Online Submission
  • NRF
  • KOFST
  • crossref crossmark
  • crossref cited-by
  • crossref funder-registry
  • crosscheck
  • orcid
  • open access
  • ccl
Journal Informaiton Journal Informaiton - close