A method for rapid field monitoring of boll opening rate for cotton based on digital imagery

doi:10.1016/j.jia.2025.07.018

Advanced Online Publication | Current Issue | Archive | Adv Search

Chenyu Xiao¹, Yukun Wang¹, Haohao Zhao¹, A. Egrinya Eneji², Dongyong Xu³, Mingwei Du¹, Xin Du⁴, Qiangzi Li⁴^#, Xiaoli Tian¹^#, Zhaohu Li¹

¹ Engineering Research Center of Plant Growth Regulator, Ministry of Education/College of Agronomy and Biotechnology, China Agricultural University, Beijing 100193, China

² Department of Soil Science, Faculty of Agriculture, Forestry and Wildlife Resources Management, University of Calabar, Nigeria

³Hebei Cottonseed Engineering Technology Research Center, Hejian 062450, China

⁴Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing 100094, China

Highlights

1. The YOLOv5 model performed well in identifying cotton bolls from digital photos.

2. Merging bounding boxes on sub-images with 400×400 pixels and 700×700 pixels could improve the detection of boundary bolls.

3. Capturing images with optimal shooting height, angle and direction achieved a R² of 0.92 in monitoring boll opening rate.

Abstract
References

Download: PDF in ScienceDirect
Export: BibTeX | EndNote (RIS)

摘要

化学脱叶催熟是棉花机械采收的前提。吐絮率是决定脱叶催熟剂施用时间、用量及收获时间的关键参数。针对人工调查吐絮率效率低、时效性差的问题，本文研究了一种基于数字图像的快速监测方法。在脱叶催熟剂喷施前7天及喷施后7、14、21天采集棉田图像，期间吐絮率范围为25%~95%。设置4个拍摄高度、5个拍摄角度和2个拍摄方向，共获取912张原始图像（单张尺寸5184×3456像素），同步调查地面实际吐絮率。将单张图像分割为500×500像素的子图像，利用4种深度学习网络识别吐絮与未吐絮棉铃，其中YOLOv5模型在识别速度与精度平衡上表现最优。针对图像分割导致的边界棉铃重复识别问题，将原始图像按10种尺寸（100、200、300、400、500、600、700、800、900、1000像素）分割，利用YOLOv5模型对各尺寸子图像中的棉铃进行识别，之后合并不同尺寸子图像中同一位置棉铃的标注框，得到校正边界框。根据与吐絮率真实值的拟合效果，确定400×400像素与700×700像素图像组合效果最佳。应用该组合比较不同拍摄参数的识别效果，发现数码相机最佳拍摄高度为冠层上方20~30 cm，拍摄角度为水平向下0~30°（吐絮率>40%）和15~30°（吐絮率<40%），拍摄方向与种植行平行。本文提出的方法可实现25%~95%吐絮率范围内的快速检测(R²>92%，rRMSE<10%)，表明其在田间应用中具有较高的精度和稳定性。

Abstract

Chemical defoliation and ripening are a prerequisite for mechanical harvesting of cotton, and the boll opening rate is a critical determinant of timing and rate of defoliates and ripening agents as well as harvest. Given the low efficiency and poor timeliness of manual determination of boll opening rates, we have developed a rapid method based on digital images. Field images were collected 7 days before and 7, 14, and 21 days after the application of harvest aids, with the boll opening rates (BOR) varying from 25 to 95%. We set four shooting heights, five shooting angles and two shooting directions, and a total of 912 original images (each 5,184×3,456 pixels) were obtained. Actual ground boll opening rates were monitored simultaneously. Each single image was segmented into 500×500 pixels sub-images. The four deep learning networks were used to identify opened and unopened cotton bolls, and YOLOv5 performed best in balancing recognition time and accuracy. To address the issue of boundary boll recognition caused by image segmentation, the original images were segmented into 10 different sizes (100, 200, 300, 400, 500, 600, 700, 800, 900, and 1,000 pixels), and YOLOv5 model was then used to identify bolls in each size of the sub-images. The bounding boxes marking cotton bolls at the same position of two different sizes of sub-images, were combined to obtain new corrected bounding boxes in merged image. Based on the true values of BOR, the best combination of sub-images is 400×400 pixels with 700×700 pixels. This combination was used to examine the recognition results of various shooting parameters, and we found that the optimal shooting height for the digital camera was 20-30 cm above the canopy, with a downward angle of 0-30° (BOR higher than 40%) and 15-30° (BOR lower than 40%) from the horizontal and shooting direction parallel to the planting rows. The method established in this study can enable a less-destructive and rapid detection of BOR in the range of 25 to 95% boll opening rate, with a model R² value >92% and a relative root mean square error <10%, suggesting its high precision and stability for field application.

Keywords: digital camera image cotton bolls recognition YOLOv5

Online: 17 July 2025

Fund:

This work was funded the China Agriculture Research System (CARS-15-16).

About author: #Correspondence Qiangzi Li, Tel: +86-10-64855094, E-mail: liqz@radi.ac.cn; Xiaoli Tian, Tel: +86-10-62734550, E-mail: tianxl@cau.edu.cn

	Service
	E-mail this article
	Add to citation manager
	E-mail Alert
	RSS
	Articles by authors

Cite this article:

Chenyu Xiao, Yukun Wang, Haohao Zhao, A. Egrinya Eneji, Dongyong Xu, Mingwei Du, Xin Du, Qiangzi Li, Xiaoli Tian, Zhaohu Li. 2025. A method for rapid field monitoring of boll opening rate for cotton based on digital imagery. Journal of Integrative Agriculture, Doi:10.1016/j.jia.2025.07.018

Dong H Z, Mao S C, Zhang W F, Chen D H. 2014. On boll-etting optimization theory for cotton cultivation and its new development. Scientia Agricultura Sinica, 47, 441-451. (in Chinese)

Dube N, Bryant B, Sari-Sarraf H, Ritchie G L. 2020. Cotton boll distribution and yield estimation using three-dimensional point cloud data. Agronomy Journal, 112, 4976-4989.

GB/T 45102-2024. 2024. Harvesting Technical Requirements for Machine-harvested Cotton. National Standard of the People’s Republic of China. (in Chinese)正文引用

Gormus O, Kurt F, El Sabagh A. 2017. Impact of defoliation timings and leaf pubescence on yield and fiber quality of cotton. Journal of Agricultural Science and Technology, 19, 903-915.

LeCun Y, Bengio Y, Hinton G. 2015. Deep Learning. Nature, 521, 436-444.

Martin A E, Collins S J, Crowe S, Girard J, Naujokaitis-Lewis I, Smith A C, Lindsay K, Mitchell S, Fahrig L. 2020. Effects of farmland heterogeneity on biodiversity are similar to-or even larger than-the effects of farming practices. Agriculture Ecosystems & Environment, 288,106698.

Murphy K M, Ludwig E, Gutierrez J, Gehan M A. 2024. Deep learning in image-based plant phenotyping. Annual Review of Plant Biology, 75, 771-795.

Redmon J, Farhadi A. 2018. YOLOv3: An Incremental Improvement. ArXiv preprint, ArXiv:1804.02767.

Ren S, He K, Girshick R, Sun J. 2017. Faster r-cnn: towards real-time object detection with region proposal networks. IEEE Transactions on Pattern Analysis and Machine Intelligence, 39, 1137-1149.

Sharma R, Sood K. 2020. Characterization of spatial variability of soil parameters in apple orchards of Himalayan region using geostatistical analysis. Communications In Soil Science and Plant Analysis, 51, 1065-1077.

Siebert J D, Stewart A M. 2006. Correlation of defoliation timing methods to optimize cotton yield, quality, and revenue. Journal of Cotton Science, 10, 146-154.

Snipes C E, Baskin C C. 1994. Influence of early defoliation on cotton yield, seed quality, and fiber properties. Field Crops Research, 37, 137-143.

Sun S, Li C, Paterson A H, Chee P W, Robertson J S. 2019. Image processing algorithms for infield single cotton boll counting and yield prediction. Computer and Electronics in Agriculture, 166, 104976

Wang S W, Zhao J Q, Cai Y C, Li Y, Qi X R, Qiu X L, Yao X, Tian Y C, Zhu Y, Cao W X, Zhang X H. 2024. A method for small-sized wheat seedlings detection: from annotation mode to model construction. Plant Methods, 20, 15

Wang X W, Wang X F, Hu C, Dai F, Xing J F, Wang E Y, Du Z H, Wang L, Guo W S. 2022. Study on the Detection of Defoliation Effect of an Improved YOLOv5x Cotton. Agriculture, 12, 1583

Woods A J, Merritt J P, Benton S A, Bolas M T. 2004. The camera convergence problem revisited. Stereoscopic Displays and Virtual Reality Systems Xi, 5291, 167-178.

Xiao S F, Fei S P, Ye Y L, Xu D M, Xie Z W, Bi K Y, Guo Y, Li B G, Zhang R, Ma Y T. 2024. 3D reconstruction and characterization of cotton bolls in situ based on UAV technology. ISPRS Journal of Photogrammetry and Remote Sensing, 209, 101-116.

Xu X, Geng Q, Gao F, Xiong D, Qiao H B, Ma X M. 2023. Segmentation and counting of wheat spike grains based on deep learning and textural feature. Plant Methods, 19, 77

Xu Z J, Latif M A, Madni S S, Rafiq A, Alam I, Habib M A. 2021. Detecting White Cotton Bolls Using High-Resolution Aerial Imagery Acquired Through Unmanned Aerial System. IEEE Access, 9, 169068-169081.

Yeom J, Jung J, Chang A, Maeda M, Landivar J. 2018. Automated open cotton boll detection for yield estimation using unmanned aircraft vehicle (UAV) data. Remote Sensing, 10, 1895

Zhang W Q, Wu S, Wen W L, Lu X J, Wang C Y, Gou W B, Li Y K, Guo X Y, Zhao C J. 2023, Three-dimensional branch segmentation and phenotype extraction of maize tassel based on deep learning. Plant Methods, 19, 76.

Zhang Y, Yang G P, Liu Y K, Wang C, Yin Y L. 2022. An improved YOLO network for unopened cotton boll detection in the field. Journal Intelligent Fuzzy System, 42, 2193-2206.

Zhou X, Wang D, Krähenbühl P. 2019. Objects as Points. ArXiv preprint, ArXiv:1904.07850.

No related articles found!

No Suggested Reading articles found!

Viewed

Full text

Abstract

Cited

Shared

Discussed

Cite this article:

About JIA

Editorial board

For authors