Panicle Ratio Network: Streamlining rice panicle measurement by deep learning with ultra-high-definition aerial images in the field

Authors: GUO, ZIYUE - Huazhong Agricultural University; Yang, Chenghai; YANG, WANGNEN - Huazhong Agricultural University; CHEN, GUOXING - Huazhong Agricultural University; JIANG, ZHAO - Huazhong Agricultural University; WANG, BOTAO - Huazhong Agricultural University; ZHANG, JIAN - Huazhong Agricultural University

Submitted to: Journal of Experimental Botany
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 6/30/2022
Publication Date: 11/2/2022
Citation: Guo, Z., Yang, C., Yang, W., Chen, G., Jiang, Z., Wang, B., Zhang, J. 2022. Panicle Ratio Network: Streamlining rice panicle measurement by deep learning with ultra-high-definition aerial images in the field. Journal of Experimental Botany. 71(19):6575-6588.

Interpretive Summary: The heading date and effective tiller percentage are important traits in rice, and they directly affect plant architecture and yield. Both traits are related to the ratio of the panicle number to the maximum tiller number, or simply called panicle ratio. In this study, an automatic estimation model based on a deep learning neural network was developed. Ultra-high-definition unmanned aerial vehicle (UAV) images were collected from cultivated rice varieties in small plots in 2019 and 2020 and in a large field in 2021. Statistical analysis showed that there existed strong correlations between estimated and ground measured panicle ratio data. Based on the result analysis, various factors affecting panicle ratio estimation and strategies for improving estimation accuracy were investigated. The results obtained in this study demonstrate the feasibility of using UAV and deep learning techniques to replace ground-based manual methods to accurately extract phenotypic information for rice and other cereal crops.

Technical Abstract: The heading date and effective tiller percentage are important traits in rice, and they directly affect plant architecture and yield. Both traits are related to the ratio of the panicle number to the maximum tiller number, or simply called panicle ratio (PR). In this study, an automatic PR estimation model (PRNet) based on a deep convolutional neural network was developed. Ultra-high-definition unmanned aerial vehicle (UAV) images were collected from cultivated rice varieties planted in 2384 experimental plots in 2019 and 2020 and in a large field in 2021. The determination coefficient between estimated PR and ground measured PR reached 0.935, and the root mean square error (RMSE) values for the estimations of the heading date and effective tiller percentage were 0.687 days and 4.84%, respectively. Based on the result analysis, various factors affecting PR estimation and strategies for improving PR estimation accuracy were investigated. The satisfactory results obtained in this study demonstrate the feasibility of using UAV and deep learning techniques to replace ground-based manual methods to accurately extract phenotypic information of crop micro targets (such as grains per panicle, panicle flowering, etc.) for rice and potentially for other cereal crops with future research.