A groundbreaking research project has developed a cutting-edge platform that combines artificial intelligence, cloud computing, and drone technology to revolutionize rice production. The platform, called Panicle-Cloud, utilizes AI-powered cloud computing to quantify rice panicles from drone-collected imagery, enabling the classification of yield production in rice.
The research team created an expert-annotated diverse rice panicle detection (DRPD) dataset and integrated multiple deep learning models into the Panicle-Cloud platform. Through iterative improvements, the team identified the Panicle AI model as the most accurate for detecting and quantifying rice panicles. By analyzing drone flights at different altitudes and growth stages, they determined that an altitude of 7m during the early grain-filling stages yielded the best results.
Correlation analysis confirmed the effectiveness of the Panicle AI model, particularly at the 7m height, with a high correlation coefficient. This model outperformed 13 other state-of-the-art deep learning models in panicle detection accuracy. The user-friendly Panicle-Cloud platform allows non-experts to select AI models for panicle detection through a web-based interface. The platform also optimizes computation by cropping larger images.
In a two-season rice field trial, the platform successfully classified rice yield production into low, medium, and high categories with an overall accuracy of over 84%. This feature enables rice breeders to effectively screen and select preferred varieties based on predicted yield performance.
By integrating AI, cloud computing, and drone technology, Panicle-Cloud offers an efficient and accurate solution for quantifying yield-related traits in rice. This technology has the potential to significantly enhance rice breeding and cultivation processes, addressing the challenges posed by climate change and increasing food demand. With its accessibility and advanced phenotyping tools, Panicle-Cloud empowers a wider range of users to select high-yielding rice varieties, contributing to food security on a global scale.
Reference: Teng, Z., et al. “Panicle-Cloud: An Open and AI-Powered Cloud Computing Platform for Quantifying Rice Panicles from Drone-Collected Imagery to Enable the Classification of Yield Production in Rice.” Plant Phenomics (2023). DOI: 10.34133/plantphenomics.0105