Recognition of Maize Seedling under Weed Disturbance using Deep Learning Algorithms DOI Creative Commons

boyi Tang,

Jingping Zhou,

Yuchun Pan

et al.

Research Square (Research Square), Journal Year: 2024, Volume and Issue: unknown

Published: March 4, 2024

Abstract Using UAV-based RGB images to recognize maize seedlings is of great significant for precise weed control, efficient water and fertilizer management. However, the presence weeds with morphological resemblances at seedling stage affects recognition seedlings. This research employs UAV deep learning algorithms achieve accurate under disturbance. Firstly, adaptive anchor frame algorithm employed intelligently select optimal sizes suited from images. strategic selection minimizes time computational demands associated multiple sampling. Subsequently, Global Attention Mechanism (GAM) introduced, bolstering feature extraction capabilities. A range models, including YOLOv3 YOLOv5, are applied recognition, culminating in identification an model. To account real-world scenarios, we investigate influences flight altitude disturbance on recognition. The results indicate a multi-class Average Precision (mAP) 94.5% 88.2% detecting altitudes 15m 30m, respectively, average detection speed 0.025s per single image. emphasizes efficacy improved YOLOv5 model recognizing using

Language: Английский

Object detection and tracking in Precision Farming: a systematic review DOI Creative Commons
Mar Ariza-Sentís, Sergio Vélez, Raquel Martínez‐Peña

et al.

Computers and Electronics in Agriculture, Journal Year: 2024, Volume and Issue: 219, P. 108757 - 108757

Published: Feb. 23, 2024

Object Detection and Tracking have gained importance in recent years because of the great advances image video analysis techniques accurate results these technologies are producing. Moreover, they successfully been applied to multiple fields, including agricultural domain since offer real-time monitoring status crops animals while counting how many present within a field/barn. This review aims current literature on field Precision Farming. For that, over 300 research articles were explored, from which 150 last five systematically reviewed analysed regarding algorithms implemented, belong to, difficulties faced, limitations should be tackled future. Lastly, it examines potential issues that this approach might have, for instance, lack open-source datasets with labelled data. The findings study indicate critical enhance Farming pave way robotization sector provide insights crop animal management, optimize resource allocation. Future work focus optimal acquisition prior Tracking, along consideration biophysical environment farming scenarios.

Language: Английский

Citations

42
Deep Learning-Based Weed–Crop Recognition for Smart Agricultural Equipment: A Review DOI Creative Commons

Hao-Ran Qu,

Wen‐Hao Su

Agronomy, Journal Year: 2024, Volume and Issue: 14(2), P. 363 - 363

Published: Feb. 11, 2024

Weeds and crops engage in a relentless battle for the same resources, leading to potential reductions crop yields increased agricultural costs. Traditional methods of weed control, such as heavy herbicide use, come with drawback promoting resistance environmental pollution. As demand pollution-free organic products rises, there is pressing need innovative solutions. The emergence smart equipment, including intelligent robots, unmanned aerial vehicles satellite technology, proves be pivotal addressing weed-related challenges. effectiveness however, hinges on accurate detection, task influenced by various factors, like growth stages, conditions shading. To achieve precise identification, it essential employ suitable sensors optimized algorithms. Deep learning plays crucial role enhancing recognition accuracy. This advancement enables targeted actions minimal pesticide spraying or laser excision weeds, effectively reducing overall cost production. paper provides thorough overview application deep equipment. Starting an tools, identification algorithms, discussion delves into instructive examples, showcasing technology’s prowess distinguishing between weeds crops. narrative highlights recent breakthroughs automated technologies precision plant while acknowledging existing challenges proposing prospects. By marrying cutting-edge technology sustainable practices, adoption equipment presents promising path toward efficient eco-friendly management modern agriculture.

Language: Английский

Citations

22
A Survey of Object Detection for UAVs Based on Deep Learning DOI Creative Commons

Guangyi Tang,

Jianjun Ni, Yonghao Zhao

et al.

Remote Sensing, Journal Year: 2023, Volume and Issue: 16(1), P. 149 - 149

Published: Dec. 29, 2023

With the rapid development of object detection technology for unmanned aerial vehicles (UAVs), it is convenient to collect data from UAV photographs. They have a wide range applications in several fields, such as monitoring, geological exploration, precision agriculture, and disaster early warning. In recent years, many methods based on artificial intelligence been proposed detection, deep learning key area this field. Significant progress has achieved deep-learning-based detection. Thus, paper presents review research This survey provides an overview UAVs summarizes UAVs. addition, issues are analyzed, small under complex backgrounds, rotation, scale change, category imbalance problems. Then, some representative solutions these summarized. Finally, future directions field discussed.

Language: Английский

Citations

35
Progress in the Application of CNN-Based Image Classification and Recognition in Whole Crop Growth Cycles DOI Creative Commons

Feng Yu,

Qian Zhang, Jun Xiao

et al.

Remote Sensing, Journal Year: 2023, Volume and Issue: 15(12), P. 2988 - 2988

Published: June 8, 2023

The categorization and identification of agricultural imagery constitute the fundamental requisites contemporary farming practices. Among various methods employed for image classification recognition, convolutional neural network (CNN) stands out as most extensively utilized swiftly advancing machine learning technique. Its immense potential precision agriculture cannot be understated. By comprehensively reviewing progress made in CNN applications throughout entire crop growth cycle, this study aims to provide an updated account these endeavors spanning years 2020 2023. During seed stage, networks are effectively categorize screen seeds. In vegetative recognition play a prominent role, with diverse range models being applied, each its own specific focus. reproductive CNN’s application primarily centers around target detection mechanized harvesting purposes. As post-harvest assumes pivotal role screening grading harvested products. Ultimately, through comprehensive analysis prevailing research landscape, presents characteristics trends current investigations, while outlining future developmental trajectory classification.

Language: Английский

Citations

26
Performance evaluation of newly released cameras for fruit detection and localization in complex kiwifruit orchard environments DOI
Xiaojuan Liu,

Xudong Jing,

Hanhui Jiang

et al.

Journal of Field Robotics, Journal Year: 2024, Volume and Issue: 41(4), P. 881 - 894

Published: Jan. 30, 2024

Abstract Consumer RGB‐D and binocular stereo cameras were applied to fruit detection localization. However, few studies are documented on performance comparison of newly released under same scene in complex orchard. This study evaluates consumer based YOLOv5x for kiwifruit localization selection optimal one with better application orchard environment. Firstly, Azure Kinect, RealSense D435, ZED 2i employed capture images canopies. Subsequently, was train detect kiwifruits calyxes the images. Meanwhile, an overlap‐partitioning strategy calyx detection. Additionally, spatial coordinate transformation performed by integrating camera's extrinsic parameters depth map generated each camera. Finally, three‐dimensional coordinates calculated compared ground truth, followed accuracy analyzed. Results show that obtained mean average precision 93.2%, 91.3%, 95.8% three detection, respectively. Overlap‐partitioning improved significantly increased 13.00%, 16.30%, 7.70%, The absolute deviation Y‐axis is relatively high at 8.44 mm 6.67 while D435 achieved minimum 10.42 X‐axis 18.33 Z‐axis. Average speed image 0.164 s, 0.037 0.062 s 2i, These results indicate excellent than Kinect orchard, which could be a valuable reference other orchards select camera capacity.

Language: Английский

Citations

10
A survey of deep learning-based object detection methods in crop counting DOI
Yuning Huang, Yurong Qian, Hongyang Wei

et al.

Computers and Electronics in Agriculture, Journal Year: 2023, Volume and Issue: 215, P. 108425 - 108425

Published: Nov. 11, 2023

Language: Английский

Citations

19
Unmanned Aerial Vehicle-Scale Weed Segmentation Method Based on Image Analysis Technology for Enhanced Accuracy of Maize Seedling Counting DOI Creative Commons
Tianle Yang, Shaolong Zhu, Weijun Zhang

et al.

Agriculture, Journal Year: 2024, Volume and Issue: 14(2), P. 175 - 175

Published: Jan. 24, 2024

The number of maize seedlings is a key determinant yield. Thus, timely, accurate estimation helps optimize and adjust field management measures. Differentiating “multiple in single hole” accurately using deep learning object detection methods presents challenges that hinder effectiveness. Multivariate regression techniques prove more suitable such cases, yet the presence weeds considerably affects accuracy. Therefore, this paper proposes weed identification method combines shape features with threshold skeleton clustering to mitigate impact on counting. (TS) ensured accuracy precision values eliminating exceeded 97% missed inspection rate misunderstanding did not exceed 6%, which significant improvement compared traditional methods. Multi-image characteristics coverage, seedling edge pixel percentage, characteristic connecting domain gradually returned seedlings. After applying TS remove weeds, estimated R2 0.83, RMSE 1.43, MAE 1.05, overall counting 99.2%. segmentation proposed can adapt various conditions. Under different emergence conditions, count reaches maximum 0.88, an below 1.29. approach study shows improved recognition drone images conventional image processing It exhibits strong adaptability stability, enhancing even weeds.

Language: Английский

Citations

7
Maize emergence rate and leaf emergence speed estimation via image detection under field rail-based phenotyping platform DOI

Lvhan Zhuang,

Chuanyu Wang,

Haoyuan Hao

et al.

Computers and Electronics in Agriculture, Journal Year: 2024, Volume and Issue: 220, P. 108838 - 108838

Published: March 19, 2024

Language: Английский

Citations

6
Rice Counting and Localization in Unmanned Aerial Vehicle Imagery Using Enhanced Feature Fusion DOI Creative Commons
Mingwei Yao, Wei Li, Li Chen

et al.

Agronomy, Journal Year: 2024, Volume and Issue: 14(4), P. 868 - 868

Published: April 21, 2024

In rice cultivation and breeding, obtaining accurate information on the quantity spatial distribution of plants is crucial. However, traditional field sampling methods can only provide rough estimates plant count fail to capture precise locations. To address these problems, this paper proposes P2PNet-EFF for counting localization plants. Firstly, through introduction enhanced feature fusion (EFF), model improves its ability integrate deep semantic while preserving shallow details. This allows holistically analyze morphology rather than focusing solely their central points, substantially reducing errors caused by leaf overlap. Secondly, integrating efficient multi-scale attention (EMA) into backbone, enhances extraction capabilities suppresses interference from similar backgrounds. Finally, evaluate effectiveness method, we introduce URCAL dataset localization, gathered using UAV. consists 365 high-resolution images 173,352 point annotations. Experimental results demonstrate that proposed method achieves a 34.87% reduction in MAE 28.19% RMSE compared original P2PNet increasing R2 3.03%. Furthermore, conducted extensive experiments three frequently used datasets. The excellent performance method.

Language: Английский

Citations

5
Lightweight Corn Leaf Detection and Counting Using Improved YOLOv8 DOI Creative Commons

Shaotong Ning,

Tan Feng, Xuebo Chen

et al.

Sensors, Journal Year: 2024, Volume and Issue: 24(16), P. 5279 - 5279

Published: Aug. 15, 2024

The number of maize leaves is an important indicator for assessing plant growth and regulating population structure. However, the traditional leaf counting method mainly relies on manual work, which both time-consuming straining, while existing image processing methods have low accuracy poor adaptability, making it difficult to meet standards practical application. To accurately detect status maize, improved lightweight YOLOv8 detection was proposed in this study. Firstly, backbone network replaced using StarNet convolution attention fusion module (CAFM) introduced, combines local global mechanisms enhance ability feature representation information from different channels. Secondly, neck part, StarBlock used improve C2f capture more complex features preserving original through jump connections training stability performance. Finally, a shared convolutional head (LSCD) reduce repetitive computations computational efficiency. experimental results show that precision, recall, mAP50 model are 97.9%, 95.5%, 97.5%, numbers parameters size 1.8 M 3.8 MB, reduced by 40.86% 39.68% compared YOLOv8. This study shows improves detection, assists breeders scientific decisions, provides reference deployment application mobile end devices, technical support high-quality assessment growth.

Language: Английский

Citations

5