EIREducational Innovation Research3029-18443029-1852WHIOCE PUBLISHING PTE. LTD.10.18063/EIR.v4i2.1569ArticleStudy on Light Response Mechanism and Yield Prediction of Tropical Dragon Fruit Integrating Multimodal Data and Machine Learninghttps://artdesignp.com/journal/EIR/4/2/10.18063/EIR.v4i2.1569JingRu,ZhuYupeng,ChenMeixia2026422026-02-26In off-season dragon fruit cultivation in Hainan, artificial supplementary lighting has become a critical method for improving winter flowering rates. However, current lighting management predominantly relies on empirical approaches, resulting in suboptimal parameters, high energy consumption, and unstable yields. To address these challenges, this study proposes a research framework integrating multimodal data and machine learning to elucidate light response mechanisms in dragon fruit and achieve precise yield prediction. The study first establishes a multimodal dataset incorporating environmental conditions, image data, and physiological indicators through field orthogonal experiments. Next, interpretable machine learning algorithms are employed to quantify nonlinear relationships between lighting parameters and flowering rates, enabling optimal lighting strategies. Subsequently, YOLO object detection combined with LSTM/Transformer time-series models facilitates automated flower-fruit tracking and dynamic yield prediction. Finally, an integrated intelligent lighting decision support system prototype is developed. This research aims to transition from experience-driven approaches to data-driven intelligent solutions, providing theoretical foundations and technical references for precision cultivation of tropical fruit trees.Pitaya, Multimodal data, Machine learning, Light response mechanism, Yield prediction, Intelligent supplementary lighting

[1] Xiong R, Xu M, Liu C, et al., 2019, Supplementary lighting conditions for inducing flowering of dragon fruit in winter in Hainan. Acta Tropical Biology, 10(1): 6. [2] Chen D, Fan W, Ou S, et al., 2019, Effects of different light quality LED lamps on light supplementation and flower induction in dragon fruit. Meteorological Research and Application, 40(2): 5. [3] Ming J, Lin X, Gao H, et al., 2020, Research progress on artificial supplementary lighting technology application in pitaya. Tropical Agricultural Science, 40(3): 6. [4] Zhang Y, Xiao W, Lu X, et al., 2021, Method for detecting rice panicle flower opening status based on visible light images. Acta Agronomica Sinica, 37(9): 11. [5] Zhu J, Xia J, Ying Y, et al., 2025, Intelligent agricultural sensors: Technological development and challenges based on application scenarios. Acta Agrarum Sinicae, 41(16): 1-12. [6] Chen Y, Li X, Wang S, et al., 2024, Integrated learning combined with multi-source data for predicting winter wheat yield per unit area in Henan Province. Journal of Agricultural Engineering, 40(4): 177-185. [7] Fei S, Yu X, Lan M, et al., 2021, Research on winter wheat yield estimation based on hyperspectral remote sensing and ensemble learning methods. China Agricultural Science, 54(16): 3417-3427. [8] Knott M, Perez-Cruz F, Defraeye T, 2023, Facilitated machine learning for image-based fruit quality assessment. Journal of Food Engineering, 345: 111401. [9] Zhao WJ, An YL, Song CQ, et al., 2025, ATR-FTIR Coupled With Machine Learning Provides a Fast Method for Identifying and Distinguishing 55 Varieties of Fruit-Derived Medicinal Materials. Phytochemical Analysis, 36(6): 1790-1802.