Research

16 pages, 4877 KiB

Open AccessArticle

Improved Method for Apple Fruit Target Detection Based on YOLOv5s

by Huaiwen Wang, Jianguo Feng and Honghuan Yin

Agriculture 2023, 13(11), 2167; https://doi.org/10.3390/agriculture13112167 - 18 Nov 2023

Cited by 5 | Viewed by 1447

Images captured using unmanned aerial vehicles (UAVs) often exhibit dense target distribution and indistinct features, which leads to the issues of missed detection and false detection in target detection tasks. To address these problems, an improved method for small target detection called YOLOv5s [...] Read more.

Images captured using unmanned aerial vehicles (UAVs) often exhibit dense target distribution and indistinct features, which leads to the issues of missed detection and false detection in target detection tasks. To address these problems, an improved method for small target detection called YOLOv5s is proposed to enhance the detection accuracy for small targets such as apple fruits. By applying improvements to the RFA module, DFP module, and Soft-NMS algorithm, as well as integrating these three modules together, accurate detection of small targets in images can be achieved. Experimental results demonstrate that the integrated, improved model achieved a significant improvement in detection accuracy, with precision, recall, and mAP increasing by 3.6%, 6.8%, and 6.1%, respectively. Furthermore, the improved method shows a faster convergence speed and lower loss value during the training process, resulting in higher recognition accuracy. The results of this study indicate that the proposed improved method exhibits a good performance in apple fruit detection tasks involving UAV imagery, which is of great significance for fruit yield estimation. The research findings demonstrate the effectiveness and feasibility of the improved method in addressing small target detection tasks, such as apple fruit detection. Full article

(This article belongs to the Special Issue Model-Assisted and Computational Plant Phenotyping)

► Show Figures

Figure 1

19 pages, 8347 KiB

Open AccessArticle

M2F-Net: A Deep Learning-Based Multimodal Classification with High-Throughput Phenotyping for Identification of Overabundance of Fertilizers

by J. Dhakshayani and B. Surendiran

Agriculture 2023, 13(6), 1238; https://doi.org/10.3390/agriculture13061238 - 13 Jun 2023

Viewed by 1881

Abstract

Amaranth, a pseudocereal crop which is rich in nutrients and climate resistant, can provide an opportunity to increase food security and nutritional content for the growing population. Farmers rely mainly on synthetic fertilizers to improve the quality and yield of the crop; however, [...] Read more.

Amaranth, a pseudocereal crop which is rich in nutrients and climate resistant, can provide an opportunity to increase food security and nutritional content for the growing population. Farmers rely mainly on synthetic fertilizers to improve the quality and yield of the crop; however, this overuse harms the ecosystem. Understanding the mechanism causing this environmental deterioration is crucial for crop production and ecological sustainability. In recent years, high-throughput phenotyping using Artificial Intelligence (AI) has been thriving and can provide an effective solution for the identification of fertilizer overuse. Influenced by the strength of deep learning paradigms and IoT sensors, a novel multimodal fusion network (M2F-Net) is proposed for high-throughput phenotyping to diagnose overabundance of fertilizers. In this paper, we developed and analyzed three strategies that fuse agrometeorological and image data by assessing fusion at various stages. Initially two unimodal baseline networks were trained: Multi-Layer Perceptron (MLP) on agrometeorological data and a pre-trained Convolutional Neural Network (CNN) model DenseNet-121 on image data. With these baselines, the multimodal fusion network is developed, capable of adeptly learning from image and non-image data and the model’s performance is evaluated in terms of accuracy and Area Under Curve (AUC). Moreover, the fusion approaches that are considered outperformed the unimodal networks remarkably with 91% accuracy. From the experimental result, it is proven that incorporating agrometeorological information and images can substantially boost the classification performance for the overabundance of fertilizer. Full article

(This article belongs to the Special Issue Model-Assisted and Computational Plant Phenotyping)

► Show Figures

Figure 1

12 pages, 24892 KiB

Open AccessArticle

Maize Stem Contour Extraction and Diameter Measurement Based on Adaptive Threshold Segmentation in Field Conditions

by Jing Zhou, Yushan Wu, Jian Chen, Mingren Cui, Yudi Gao, Keying Meng, Min Wu, Xinyu Guo and Weiliang Wen

Agriculture 2023, 13(3), 678; https://doi.org/10.3390/agriculture13030678 - 14 Mar 2023

Cited by 4 | Viewed by 1801

Abstract

Solving the problem of the stem contour extraction of maize is difficult under open field conditions, and the stem diameter cannot be measured quickly and nondestructively. In this paper, maize at the small and large bell stages was the object of study. An [...] Read more.

Solving the problem of the stem contour extraction of maize is difficult under open field conditions, and the stem diameter cannot be measured quickly and nondestructively. In this paper, maize at the small and large bell stages was the object of study. An adaptive threshold segmentation algorithm based on the color space model was proposed to obtain the stem contour and stem diameter of maize in the field. Firstly, 2D images of the maize stem in the field were captured with an RGB-D camera. Then, the images were processed by hue saturation value (HSV) color space. Next, the stem contour of the maize was extracted by maximum between-class variance (Otsu). Finally, the reference method was used to obtain the stem diameter of the maize. Scatter plots and Dice coefficients were used to compare the contour extraction effects of the HSV + fixed threshold algorithm, the HSV + Otsu algorithm, and the HSV + K-means algorithm. The results showed that the HSV + Otsu algorithm is the optimal choice for extracting the maize stem contour. The mean absolute error, mean absolute percentage error (MAPE), and root mean square error (RMSE) of the maize stem diameter at the small bell stage were 4.30 mm, 10.76%, and 5.29 mm, respectively. The mean absolute error, MAPE, and RMSE of the stem diameter of the maize at the large bell stage were 4.78 mm, 12.82%, and 5.48 mm, respectively. The MAPE was within 10–20%. The results showed that the HSV + Otsu algorithm could meet the requirements for stem diameter measurement and provide a reference for the acquisition of maize phenotypic parameters in the field. In the meantime, the acquisition of maize phenotypic parameters under open field conditions provides technical and data support for precision farming and plant breeding. Full article

(This article belongs to the Special Issue Model-Assisted and Computational Plant Phenotyping)

► Show Figures

Figure 1

19 pages, 2023 KiB

Open AccessFeature PaperArticle

Precise Recommendation Method of Suitable Planting Areas of Maize Varieties Based on Knowledge Graph

by Yidong Zou, Shouhui Pan, Feng Yang, Dongfeng Zhang, Yanyun Han, Xiangyu Zhao, Kaiyi Wang and Chunjiang Zhao

Agriculture 2023, 13(3), 526; https://doi.org/10.3390/agriculture13030526 - 22 Feb 2023

Cited by 2 | Viewed by 1174

Abstract

The rapid increase in the number of new maize varieties and the intensification of market competition have raised the need to precisely promote new maize varieties to suitable planting areas and fully exploit the variety potential and win the market competition. This paper [...] Read more.

The rapid increase in the number of new maize varieties and the intensification of market competition have raised the need to precisely promote new maize varieties to suitable planting areas and fully exploit the variety potential and win the market competition. This paper proposes a precise recommendation method for suitable planting areas of maize varieties based on a knowledge graph. The meteorology knowledge graph of maize ecological regions is constructed at county-scale and a RippleNet recommendation model is used to mine the potential spatial correlation of maize variety suitability in different meteorological environments. The county-scale precise recommendation for suitable planting areas is then realized. In total, 331 maize varieties and agricultural meteorological data of 59 experimental areas in the Huang-Huai-Hai ecological region are used for model training and testing (accuracy 76.3%). Through experimental comparison, the recommendation accuracy of this method is 24.3% higher than that of six traditional machine learning methods, 11.2% higher than that of graph attention networks, and 5.8% higher than that of graph convolution neural networks. This study provides a data-driven solution for the precise recommendation and market positioning of maize varieties, enhances the scientificity of variety recommendation and helps to fully exploit their planting potential. Full article

(This article belongs to the Special Issue Model-Assisted and Computational Plant Phenotyping)

► Show Figures

Figure 1

17 pages, 5362 KiB

Open AccessArticle

Crop Node Detection and Internode Length Estimation Using an Improved YOLOv5 Model

by Jinnan Hu, Guo Li, Haolan Mo, Yibo Lv, Tingting Qian, Ming Chen and Shenglian Lu

Agriculture 2023, 13(2), 473; https://doi.org/10.3390/agriculture13020473 - 16 Feb 2023

Cited by 5 | Viewed by 2358

Abstract

The extraction and analysis of plant phenotypic characteristics are critical issues for many precision agriculture applications. An improved YOLOv5 model was proposed in this study for accurate node detection and internode length estimation of crops by using an end-to-end approach. In this improved [...] Read more.

The extraction and analysis of plant phenotypic characteristics are critical issues for many precision agriculture applications. An improved YOLOv5 model was proposed in this study for accurate node detection and internode length estimation of crops by using an end-to-end approach. In this improved YOLOv5, a feature extraction module was added in front of each detection head, and the bounding box loss function used in the original network of YOLOv5 was replaced by the SIoU bounding box loss function. The results of the experiments on three different crops (chili, eggplant, and tomato) showed that the improved YOLOv5 reached 90.5% AP (average precision) and the average detection time was 0.019 s per image. The average error of the internode length estimation was 41.3 pixels, and the relative error was 7.36%. Compared with the original YOLOv5, the improved YOLOv5 had an average error reduction of 5.84 pixels and a relative error reduction of 1.61%. Full article

(This article belongs to the Special Issue Model-Assisted and Computational Plant Phenotyping)

► Show Figures

Figure 1

19 pages, 63698 KiB

Open AccessArticle

Evaluating Data Augmentation Effects on the Recognition of Sugarcane Leaf Spot

by Yiqi Huang, Ruqi Li, Xiaotong Wei, Zhen Wang, Tianbei Ge and Xi Qiao

Agriculture 2022, 12(12), 1997; https://doi.org/10.3390/agriculture12121997 - 24 Nov 2022

Cited by 6 | Viewed by 1695

Abstract

Research on the recognition and segmentation of plant diseases in simple environments based on deep learning has achieved relative success. However, under the conditions of a complex environment and a lack of samples, the model has difficulty recognizing disease spots, or its recognition [...] Read more.

Research on the recognition and segmentation of plant diseases in simple environments based on deep learning has achieved relative success. However, under the conditions of a complex environment and a lack of samples, the model has difficulty recognizing disease spots, or its recognition accuracy is too low. This paper is aimed at investigating how to improve the recognition accuracy of the model when the dataset is in a complex environment and lacks samples. First, for the complex environment, this paper uses DeepLabV3+ to segment sugarcane leaves from complex backgrounds; second, focusing on the lack of training images of sugarcane leaves, two data augmentation methods are used in this paper: supervised data augmentation and deep convolutional generative adversarial networks (DCGANs) for data augmentation. MobileNetV3-large, Alexnet, Resnet, and Densenet are trained by comparing the original dataset, original dataset with supervised data augmentation, original dataset with DCGAN augmentation, background-removed dataset, background-removed dataset with supervised data augmentation, and background-removed dataset with DCGAN augmentation. Then, the recognition abilities of the trained models are compared using the same test set. The optimal network selected based on accuracy and training time is MobileNetV3-large. Classification using MobileNetV3-large trained by the original dataset yielded 53.5% accuracy. By removing the background and adding synthetic images produced by the DCGAN, the accuracy increased to 99%. Full article

(This article belongs to the Special Issue Model-Assisted and Computational Plant Phenotyping)

► Show Figures

Figure 1

18 pages, 3106 KiB

Open AccessArticle

The Application of Machine Learning Models Based on Leaf Spectral Reflectance for Estimating the Nitrogen Nutrient Index in Maize

by Bo Chen, Xianju Lu, Shuan Yu, Shenghao Gu, Guanmin Huang, Xinyu Guo and Chunjiang Zhao

Agriculture 2022, 12(11), 1839; https://doi.org/10.3390/agriculture12111839 - 02 Nov 2022

Cited by 4 | Viewed by 1961

Abstract

Non-destructive acquisition and accurate real-time assessment of nitrogen (N) nutritional status are crucial for nitrogen management and yield prediction in maize production. The objective of this study was to develop a method for estimating the nitrogen nutrient index (NNI) of maize using in [...] Read more.

Non-destructive acquisition and accurate real-time assessment of nitrogen (N) nutritional status are crucial for nitrogen management and yield prediction in maize production. The objective of this study was to develop a method for estimating the nitrogen nutrient index (NNI) of maize using in situ leaf spectroscopy. Field trials with six nitrogen fertilizer levels (0, 75, 150, 225, 300, and 375 kg N ha⁻¹) were performed using eight summer maize cultivars. The leaf reflectance spectrum was acquired at different growth stages, with simultaneous measurements of leaf nitrogen content (LNC) and leaf dry matter (LDW). The competitive adaptive reweighted sampling (CARS) algorithm was used to screen the raw spectrum’s effective bands related to the NNI during the maize critical growth period (from the 12th fully expanded leaf stage to the milk ripening stage). Three machine learning methods—partial least squares (PLS), artificial neural networks (ANN), and support vector machines (SVM)—were used to validate the NNI estimation model. These methods indicated that the NNI first increased and then decreased (from the 12th fully expanded leaf stage to the milk ripening stage) and was positively correlated with nitrogen application. The results showed that combining effective bands and PLS (CARS-PLS) achieved the best model for NNI estimation, which yielded the highest coefficient of determination (R²_val), 0.925, and the lowest root mean square error (RMSE_val), 0.068, followed by the CARS-SVM model (R²_val, 0.895; RMSE_val, 0.081), and the CARS-ANN model (R²_val, 0.814; RMSE_val, 0.108), which performed the worst. The CARS-PLS model was used to successfully predict the variation in the NNI among cultivars and different growth stages. The estimated R² of eight cultivars by the NNI was between 0.86 and 0.97; the estimated R² of the NNI at different growth stages was between 0.92 and 0.94. The overall results indicated that the CARS-PLS allows for rapid, accurate, and non-destructive estimation of the NNI during maize growth, providing an efficient tool for accurately monitoring nitrogen nutrition. Full article

(This article belongs to the Special Issue Model-Assisted and Computational Plant Phenotyping)

► Show Figures

Figure 1

15 pages, 4023 KiB

Open AccessArticle

An Enhanced YOLOv5 Model for Greenhouse Cucumber Fruit Recognition Based on Color Space Features

by Ning Wang, Tingting Qian, Juan Yang, Linyi Li, Yingyu Zhang, Xiuguo Zheng, Yeying Xu, Hanqing Zhao and Jingyin Zhao

Agriculture 2022, 12(10), 1556; https://doi.org/10.3390/agriculture12101556 - 27 Sep 2022

Cited by 13 | Viewed by 2425

Abstract

The identification of cucumber fruit is an essential procedure in automated harvesting in greenhouses. In order to enhance the identification ability of object detection models for cucumber fruit harvesting, an extended RGB image dataset (n = 801) with 3943 positive and negative [...] Read more.

The identification of cucumber fruit is an essential procedure in automated harvesting in greenhouses. In order to enhance the identification ability of object detection models for cucumber fruit harvesting, an extended RGB image dataset (n = 801) with 3943 positive and negative labels was constructed. Firstly, twelve channels in four color spaces (RGB, YCbCr, HIS, La*b*) were compared through the ReliefF method to choose the channel with the highest weight. Secondly, the RGB image dataset was converted to the pseudo-color dataset of the chosen channel (Cr channel) to pre-train the YOLOv5s model before formal training using the RGB image dataset. Based on this method, the YOLOv5s model was enhanced by the Cr channel. The experimental results show that the cucumber fruit recognition precision of the enhanced YOLOv5s model was increased from 83.7% to 85.19%. Compared with the original YOLOv5s model, the average values of AP, F1, recall rate, and mAP were increased by 8.03%, 7%, 8.7%, and 8%, respectively. In order to verify the applicability of the pre-training method, ablation experiments were conducted on SSD, Faster R-CNN, and four YOLOv5 versions (s, l, m, x), resulting in the accuracy increasing by 1.51%, 3.09%, 1.49%, 0.63%, 3.15%, and 2.43%, respectively. The results of this study indicate that the Cr channel pre-training method is promising in enhancing cucumber fruit detection in a near-color background. Full article

(This article belongs to the Special Issue Model-Assisted and Computational Plant Phenotyping)

► Show Figures

Figure 1

Journal Menu

Journal Browser

Model-Assisted and Computational Plant Phenotyping

Share This Special Issue

Special Issue Editors

Special Issue Information

Keywords

Published Papers (8 papers)

Research

Further Information

Guidelines

MDPI Initiatives

Follow MDPI