无人机影像的玉米植株中心检测模型和方法

doi:10.3778/j.issn.1002-8331.2403-0447

摘要/Abstract

摘要： 为了解决无人机航拍图片玉米植株中心检测所面临的诸多挑战，包括植株遮挡、形态多样、光照变化以及视觉混淆等问题，提升检测精度和模型的鲁棒性，开发了一种基于YOLO-TSCAS（YOLO with triplet-attention，saliency-adaptive，and centroid awareness for scenes）模型的玉米植株中心检测算法。该算法通过三重注意力模块、显著性裁剪混合数据增强方法、自适应池化技术和选择性多单元激活函数等技术手段，有效提高了检测精度和鲁棒性。采用三重注意力模块解决目标遮挡和视觉混淆问题，使模型能够更好地关注植株中心区域。采用显著性裁剪混合数据增强方法，在训练过程中引入不同的环境和光照变化，增强了模型对复杂场景的适应能力。结合自适应池化技术提高空间分辨率，有助于捕捉更精细的特征信息，提高检测的准确性。采用选择性多单元激活函数进一步增强了网络的学习能力，使模型能够更好地适应各种场景下的植株中心检测任务。实验结果表明，与现有的YOLOX算法相比，YOLO-TSCAS算法在平均准确率和平均F1值上分别提高了22.73个百分点和0.255，并且平均对数漏检率也显著降低了0.35。与其他流行的检测模型相比，在两类植株中心目标检测精度上也取得了最佳效果。

关键词: 中心检测, 三重注意力模块, 显著性裁剪混合, 自适应池化技术, 选择性多单元

Abstract: In order to solve many challenges faced by maize plant center detection in UAV aerial images, including plant occlusion, morphological diversity, illumination change and visual confusion, and improve detection accuracy and robustness of the model, a maize plant center detection algorithm based on YOLO-TSCAS (YOLO with triplet-attention, saliency-adaptive, and centroid awareness for scenes) model is developed. The algorithm effectively improves the detection accuracy and robustness by means of triple attention module, Saliency_CutMix data enhancement method, adaptive pooling technology and selective multi-unit activation function. The triple attention module is used to solve the problems of object occlusion and visual confusion, so that the model can better focus on the central area of the plant. The Saliency_CutMix data enhancement method is used to introduce different environment and illumination changes during the training process, which increases the adaptability of the model to complex scenes. Combined with adaptive pooling technology, the spatial resolution is improved, which helps to capture finer feature information and improve the accuracy of detection. The selective multi-unit activation function is used to further enhance the learning ability of the network, so that the model can better adapt to the plant center detection task in various scenarios. The experimental results show that compared with the existing YOLOX algorithm, the YOLO-TSCAS algorithm improves the average accuracy and average F1 value by 22.73 percentage points and 0.255 respectively, and the average logarithmic missed detection rate is also significantly reduced by 0.35. Compared with other popular detection models, the best results are achieved in the detection accuracy of the two types of plant center objects.

Key words: center detection, triple attention module, Saliency_CutMix, adaptive pooling technology, selective multi-unit

邬开俊, 白晨帅, 杜建军, 张红娜, 白晓凤. 无人机影像的玉米植株中心检测模型和方法[J]. 计算机工程与应用, 2025, 61(16): 324-336.

WU Kaijun, BAI Chenshuai, DU Jianjun, ZHANG Hongna, BAI Xiaofeng. Model and Method of Maize Plant Center Detection Based on UAV Image[J]. Computer Engineering and Applications, 2025, 61(16): 324-336.

参考文献

[1] OLIVEIRA R C, SILVA R D S. Artificial intelligence in agriculture: benefits, challenges, and trends[J]. Applied Sciences, 2023, 13(13): 7405.
[2] CHEN F Q, JI X Z, BAI M X, et al. Network analysis of different exogenous hormones on the regulation of deep sowing tolerance in maize seedlings[J]. Frontiers in Plant Science, 2021, 12: 739101.
[3] FAN J H, ZHOU J, WANG B W, et al. Estimation of maize yield and flowering time using multi-temporal UAV-based hyperspectral data[J]. Remote Sensing, 2022, 14(13): 3052.
[4] CHEN S, LIU W H, FENG P Y, et al. Improving spatial disaggregation of crop yield by incorporating machine learning with multisource data: a case study of Chinese maize yield[J]. Remote Sensing, 2022, 14(10): 2340.
[5] YANG Z R, CAO Y B, SHI Y T, et al. Genetic and molecular exploration of maize environmental stress resilience: toward sustainable agriculture[J]. Molecular Plant, 2023, 16(10): 1496-1517.
[6] WU X Y, TONG L, KANG S Z, et al. Combination of suitable planting density and nitrogen rate for high yield maize and their source-sink relationship in Northwest China[J]. Journal of the Science of Food and Agriculture, 2023, 103(11): 5300-5311.
[7] SHEN J N, WANG Q L, ZHAO M, et al. Mapping maize planting densities using unmanned aerial vehicles, multispectral remote sensing, and deep learning technology[J]. Drones, 2024, 8(4): 140.
[8] QIAN B X, HUANG W J, XIE D H, et al. Coupled maize model: a 4D maize growth model based on growing degree days[J]. Computers and Electronics in Agriculture, 2023, 212: 108124.
[9] CHANG Y B, LATHAM J, LICHT M, et al. A data-driven crop model for maize yield prediction[J]. Communications Biology, 2023, 6: 439.
[10] RUIZ A, TRIFUNOVIC S, EUDY D M, et al. Harvest index has increased over the last 50 years of maize breeding[J]. Field Crops Research, 2023, 300: 108991.
[11] SZéLES A, HORVáTH é, SIMON K, et al. Maize production under drought stress: nutrient supply, yield prediction[J]. Plants, 2023, 12(18): 3301.
[12] KIENBAUM L, CORREA ABONDANO M, BLAS R, et al. DeepCob: precise and high-throughput analysis of maize cob geometry using deep learning with an application in genebank phenomics[J]. Plant Methods, 2021, 17(1): 91.
[13] ZHANG W L, WANG J Q, LIU Y X, et al. Deep-learning-based in-field citrus fruit detection and tracking[J]. Horticulture Research, 2022, 9: uhac003.
[14] WEN C J, WU J S, CHEN H R, et al. Wheat spike detection and counting in the field based on SpikeRetinaNet[J]. Frontiers in Plant Science, 2022, 13: 821717.
[15] WANG H J, LIN Y Y, XU X J, et al. A study on long-close distance coordination control strategy for litchi picking[J]. Agronomy, 2022, 12(7): 1520.
[16] TANG Y C, ZHOU H, WANG H J, et al. Fruit detection and positioning technology for a Camellia oleifera C. Abel orchard based on improved YOLOv4-tiny model and binocular stereo vision[J]. Expert Systems with Applications, 2023, 211: 118573.
[17] WU C X, LUO J Y, XIAO Y J. Multi-omics assists genomic prediction of maize yield with machine learning approaches[J]. Molecular Breeding, 2024, 44(2): 14.
[18] DE SáLEIT?O D A H, SHARMA A K, SINGH A, et al. Yield and plant height predictions of irrigated maize through unmanned aerial vehicle in North Florida[J]. Computers and Electronics in Agriculture, 2023, 215: 108374.
[19] KANG J, LIU L T, ZHANG F C, et al. Semantic segmentation model of cotton roots in situ image based on attention mechanism[J]. Computers and Electronics in Agriculture, 2021, 189: 106370.
[20] MEI W Y, WANG H Y, FOUHEY D, et al. Using deep learning and very-high-resolution imagery to map smallholder field boundaries[J]. Remote Sensing, 2022, 14(13): 3046.
[21] XU H, BLONDER B, JODRA M, et al. Automated and accurate segmentation of leaf venation networks via deep learning[J]. New Phytologist, 2021, 229(1): 631-648.
[22] YANG S, ZHENG L H, YANG H J, et al. A synthetic datasets based instance segmentation network for high-throughput soybean pods phenotype investigation[J]. Expert Systems with Applications, 2022, 192: 116403.
[23] GE Z, LIU S T, WANG F, et al. YOLOX: exceeding YOLO series in 2021[J]. arXiv:2107.08430, 2021.
[24] ZHU X K, LYU S C, WANG X, et al. TPH-YOLOv5: improved YOLOv5 based on transformer prediction head for object detection on drone-captured scenarios[C]//Proceedings of the 2021 IEEE/CVF International Conference on Computer Vision Workshops. Piscataway: IEEE, 2021: 2778-2788.
[25] LI C, LI L, JIANG H, et al. YOLOv6: a single-stage object detection framework for industrial applications[J]. arXiv:2209.02976, 2022.
[26] WANG C Y, BOCHKOVSKIY A, LIAO H M. YOLOv7: trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[C]//Proceedings of the 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2023: 7464-7475.
[27] LI Y T, FAN Q S, HUANG H S, et al. A modified YOLOv8 detection network for UAV aerial image recognition[J]. Drones, 2023, 7(5): 304.
[28] 张晓勐, 朱德利, 余茂生. 无人机遥感图像中的玉米雄穗轻量化检测模型[J]. 江西农业大学学报, 2022, 44(2): 461-472.
ZHANG X M, ZHU D L, YU M S. Lightweight detection model of maize tassel in UAV remote sensing image[J]. Acta Agriculturae Universitatis Jiangxiensis, 2022, 44(2): 461-472.
[29] 杨蜀秦, 刘江川, 徐可可, 等. 基于改进CenterNet的玉米雄蕊无人机遥感图像识别[J]. 农业机械学报, 2021, 52(9): 206-212.
YANG S Q, LIU J C, XU K K, et al. Improved CenterNet based maize tassel recognition for UAV remote sensing image[J]. Transactions of the Chinese Society for Agricultural Machinery, 2021, 52(9): 206-212.
[30] XU B R, CHAI L, ZHANG C L. Research and application on corn crop identification and positioning method based on machine vision[J]. Information Processing in Agriculture, 2023, 10(1): 106-113.
[31] HUQ A, ZERMAS D, BEBIS G. Identification of abnormality in maize plants from UAV images using deep learning approaches[C]//Proceedings of the 18th International Symposium on Visual Computing. Cham: Springer, 2023: 583-596.
[32] 梁胤豪, 陈全, 董彩霞, 等. 基于深度学习和无人机遥感技术的玉米雄穗检测研究[J]. 福建农业学报, 2020, 35(4): 456-464.
LIANG Y H, CHEN Q, DONG C X, et al. Application of deep-learning and UAV for field surveying corn tassel[J]. Fujian Journal of Agricultural Sciences, 2020, 35(4): 456-464.
[33] MOTA-DELFIN C, LóPEZ-CANTE?S G J, LóPEZ-CRUZ I L, et al. Detection and counting of corn plants in the presence of weeds with convolutional neural networks[J]. Remote Sensing, 2022, 14(19): 4892.
[34] KITANO B T, MENDES C C T, GEUS A R, et al. Corn plant counting using deep learning and UAV images[J]. IEEE Geoscience and Remote Sensing Letters, 2019. DOI:10.1109/LGRS.2019.2930549.
[35] LIU Y L, CEN C J, CHE Y P, et al. Detection of maize tassels from UAV RGB imagery with Faster R-CNN[J]. Remote Sensing, 2020, 12(2): 338.
[36] 袁亚哲. 基于无人机图像的玉米植株识别技术研究[D]. 郑州: 河南农业大学, 2019.
YUAN Y Z. Maize plant recognition technology based on UAV images[D]. Zhengzhou: Henan Agricultural University, 2019.
[37] 冯浩, 杨祯婷, 陈浩, 等. 基于无人机多光谱影像的夏玉米SPAD估算模型研究[J]. 农业机械学报, 2022, 53(10): 211-219.
FENG H, YANG Z T, CHEN H, et al. Estimation of summer maize SPAD based on UAV multispectral images[J]. Transactions of the Chinese Society for Agricultural Machinery, 2022, 53(10): 211-219.
[38] 高嘉欣. 基于无人机图像的玉米漏苗自动检测方法研究[D]. 大庆: 黑龙江八一农垦大学, 2023.
GAO J X. Research on automatic detection method of maize seedling leakage based on UAV image[D]. Daqing: Heilongjiang Bayi Agricultural University, 2023.
[39] 张新龙. 基于无人机影像的玉米出苗质量评价方法研究[D]. 石河子: 石河子大学, 2022.
ZHANG X L. Study on evaluation method of maize seedling emergence quality based on UAV image[D]. Shihezi: Shihezi University, 2022.
[40] 冯文斌. 基于无人机多光谱遥感的夏玉米长势监测及产量估测[D]. 泰安: 山东农业大学, 2022.
FENG W B. Summer corn growth monitoring and yield estimation by UAV-based multispectral remote sensing[D]. Tai’an: Shandong Agricultural University, 2022.
[41] MISRA D, NALAMADA T, ARASANIPALAI A U, et al. Rotate to attend: convolutional triplet attention module[C]//Proceedings of the 2021 IEEE Winter Conference on Applications of Computer Vision. Piscataway: IEEE, 2021: 3138-3147.
[42] UDDIN A F M, MONIRA M S, SHIN W, et al. SaliencyMix: a saliency guided data augmentation strategy for better regularization[J]. arXiv:2006.01791, 2020.
[43] STERGIOU A, POPPE R. AdaPool: exponential adaptive pooling for information-retaining downsampling[J]. IEEE Transactions on Image Processing, 2022, 32: 251-266.
[44] BISWAS K, KUMAR S, BANERJEE S, et al. SMU: smooth activation function for deep networks using smoothing maximum technique[J]. arXiv:2111.04682, 2021.
[45] HENDRYCKS D, GIMPEL K. Gaussian error linear units (GELUs)[J]. arXiv:1606.08415, 2016.