Sign in Subscribe
Machine learning

Advancements in UAV Multispectral Systems Enhance Yield Estimation for Japonica Rice in Cold Regions

Strategic Insights

— 1 min read

Researchers have made significant strides in unmanned aerial vehicle (UAV) multispectral systems, enabling the precise and efficient estimation of japonica rice yield in cold regions within the framework of precision agriculture. This innovation presents a viable alternative to conventional yield estimation methods and holds substantial practical value for rice yield estimation and sustainable rice production. By integrating UAV multispectral data with machine learning techniques, researchers can derive critical phenotypic parameters of rice and estimate yield with high accuracy.

Key Takeaways:

  • The study investigated six fertilization gradient experiments involving two conventional japonica rice varieties (KY131, SJ22) and two hybrid japonica rice varieties (CY31, TLY619) at Yanjiagang Farm in Heilongjiang Province during 2023.
  • The research integrated UAV multispectral data with machine learning techniques, including Random Forest (RF), XGBoost, Support Vector Regression (SVR), and Backpropagation Neural Network (BPNN), to derive critical phenotypic parameters of rice and estimate yield.
  • The findings revealed that phenotypic traits at critical growth stages exhibited a strong correlation with rice yield, with correlation coefficients for leaf area index (LAI) and canopy cover (CC) exceeding 0.85.
  • The accuracy of phenotypic trait evaluation using multispectral data was high, with R2 values for CC, PH, and LAI using machine learning algorithms exceeding 0.8.
  • Yield estimation performance was optimal at the heading (HD) stage, with the RF model achieving superior accuracy (R2 = 0.86, RMSE = 0.59 t/ha) compared to other growth stages.

Statistics:

  • The study involved six fertilization gradient experiments with two conventional japonica rice varieties and two hybrid japonica rice varieties.
  • The research integrated UAV multispectral data with machine learning techniques to derive critical phenotypic parameters of rice and estimate yield.
  • The correlation coefficients for LAI and CC exceeded 0.85, indicating a strong correlation with rice yield.
  • R2 values for CC, PH, and LAI using machine learning algorithms exceeded 0.8, demonstrating high accuracy in phenotypic trait evaluation.
  • The R2 for CC based on the RF algorithm exceeded 0.9, while R2 values for PH and AGB using the RF algorithm and for LAI using the XGBoost algorithm all surpassed 0.8.

Sources:

  • Machine Learning Models for Yield Estimation of Hybrid and Conventional Japonica Rice Cultivars Using Uav Imagery. Sustainability, 2025;17(18):8515.

Read more