Artificial Intelligence in Agriculture最新文献

In the uncertainties within which the worldwide food security lies nowadays, the agricultural industry is raising a crucial need for being equipped with the state-of-the-art technologies for a more efficient, climate-resilient and sustainable production. The traditional production methods have to be revisited, and opportunities should be given for the innovative solutions henceforth brought by big data analytics, cloud computing and internet of things (IoT). In this context, we develop an optimized tinyML-oriented model for an active machine learning-based greenhouse microclimate management to be integrated in an on-field microcontroller. We design an experimental strawberry greenhouse from which we collect multivariate climate data through installed sensors. The obtained values' combinations are labeled according to a five-action multi-label control strategy, then used to prepare a machine learning-ready dataset. The dataset is used to train and five-fold cross-validate 90 Multi-Layer Perceptrons (MLPs) with varied hyperparameters to select the most performant –yet optimized– model instance for the addressed task. Our multi-label control approach enables designing highly scalable models with reduced computational complexity, comprising only n control neurons instead of (1 + ∑_n^k=1C_n^k) neurons (usually generated from a classic single-label approach from n input variables). Our final selected model incorporates 2 hidden layers with 7 and 8 neurons respectively and 151 parameters; it scored a mean accuracy of 97% during the cross-validation phase, then 96% on our supplementary test set. The model enables an intelligent and autonomous greenhouse management with the less required computations. It can be efficiently deployed in microcontrollers within real world operating conditions.

Degradation of vegetation cover and expansion of barren land are remained the leading environmental problem at global level. Land surface temperature (LST), Normalized Difference Vegetation Index (NDVI), Normalized Difference Barren Index (NDBaI), and Modified Normalized Difference Water Index (MNDWI) were used to quantify the changing relationships using correlation analysis. This study attempted to analyze the relationship between LST and NDVI, NDBaI, and MNDWI using Geospatial technologies in Gida Kiremu, Limu, and Amuru districts in Western Ethiopia. All indices were estimated by using thermal bands and multispectral bands from Landsat TM 1990, Landsat ETM+ 2003, and Landsat OLI/TIRS 2020. The correlation of LST with NDVI, NDBaI and MNDWI were analyzed by using scatter plot. Accordingly, the NDBaI was positive correlation with LST (R² = 0.96). However, NDVI and MNDWI were substantially negative relationship with LST (R² = 0.99, 0.95), respectively. The result shows that, LST was increased by 5 °C due to decline of vegetation cover and increasing of bare land over the study periods. Finally, our result recommended that, decision-makers and environmental analysts should give attention on the importance of vegetation cover, water bodies and wetland in climate change mitigation, particularly, LST in the study area.

The agriculture industry is undergoing a rapid digital transformation and is growing powerful by the pillars of cutting-edge approaches like artificial intelligence and allied technologies. At the core of artificial intelligence, deep learning-based computer vision enables various agriculture activities to be performed automatically with utmost precision enabling smart agriculture into reality. Computer vision techniques, in conjunction with high-quality image acquisition using remote cameras, enable non-contact and efficient technology-driven solutions in agriculture. This review contributes to providing state-of-the-art computer vision technologies based on deep learning that can assist farmers in operations starting from land preparation to harvesting. Recent works in the area of computer vision were analyzed in this paper and categorized into (a) seed quality analysis, (b) soil analysis, (c) irrigation water management, (d) plant health analysis, (e) weed management (f) livestock management and (g) yield estimation. The paper also discusses recent trends in computer vision such as generative adversarial networks (GAN), vision transformers (ViT) and other popular deep learning architectures. Additionally, this study pinpoints the challenges in implementing the solutions in the farmer’s field in real-time. The overall finding indicates that convolutional neural networks are the corner stone of modern computer vision approaches and their various architectures provide high-quality solutions across various agriculture activities in terms of precision and accuracy. However, the success of the computer vision approach lies in building the model on a quality dataset and providing real-time solutions.

Point-cloud data acquired using a terrestrial laser scanner play an important role in digital forestry research. Multiple scans are generally used to overcome occlusion effects and obtain complete tree structural information. However, the placement of artificial reflectors in a forest with complex terrain for marker-based registration is time-consuming and difficult. In this study, an automatic coarse-to-fine method for the registration of point-cloud data from multiple scans of a single tree was proposed. In coarse registration, point clouds produced by each scan are projected onto a spherical surface to generate a series of two-dimensional (2D) images, which are used to estimate the initial positions of multiple scans. Corresponding feature-point pairs are then extracted from these series of 2D images. In fine registration, point-cloud data slicing and fitting methods are used to extract corresponding central stem and branch centers for use as tie points to calculate fine transformation parameters. To evaluate the accuracy of registration results, we propose a model of error evaluation via calculating the distances between center points from corresponding branches in adjacent scans. For accurate evaluation, we conducted experiments on two simulated trees and six real-world trees. Average registration errors of the proposed method were 0.026 m around on simulated tree point clouds, and 0.049 m around on real-world tree point clouds.

Sericulture is the process of cultivating silkworms for the production of silk. High-quality production of silk without mixing with low quality is a great challenge faced in the silk production centers. One of the possibilities to overcome this issue is by separating male and female cocoons before extracting silk fibers from the cocoons as male cocoon silk fibers are finer than females. This study proposes a method for the classification of male and female cocoons with the help of X-ray images without destructing the cocoon. The study used popular single hybrid varieties FC1 and FC2 mulberry silkworm cocoons. The shape features of the pupa are considered for the classification process and were obtained without cutting the cocoon. A novel point interpolation method is used for the computation of the width and height of the cocoon. Different dimensionality reduction methods are employed to enhance the performance of the model. The preprocessed features are fed to the powerful ensemble learning method AdaBoost and used logistic regression as the base learner. This model attained a mean accuracy of 96.3% for FC1 and FC2 in cross-validation and 95.3% in FC1 and 95.1% in FC2 for external validation.

The feral or volunteer cotton (VC) plants when reach the pinhead squaring phase (5–6 leaf stage) can act as hosts for the boll weevil (Anthonomus grandis L.) pests. The Texas Boll Weevil Eradication Program (TBWEP) employs people to locate and eliminate VC plants growing by the side of roads or fields with rotation crops but the ones growing in the middle of fields remain undetected. In this paper, we demonstrate the application of computer vision (CV) algorithm based on You Only Look Once version 5 (YOLOv5) for detecting VC plants growing in the middle of corn fields at three different growth stages (V3, V6 and VT) using unmanned aircraft systems (UAS) remote sensing imagery. All the four variants of YOLOv5 (s, m, l, and x) were used and their performances were compared based on classification accuracy, mean average precision (mAP) and F1-score. It was found that YOLOv5s could detect VC plants with maximum classification accuracy of 98% and mAP of 96.3% at V6 stage of corn while YOLOv5s and YOLOv5m resulted in the lowest classification accuracy of 85% and YOLOv5m and YOLOv5l had the least mAP of 86.5% at VT stage on images of size 416 × 416 pixels. The developed CV algorithm has the potential to effectively detect and locate VC plants growing in the middle of corn fields as well as expedite the management aspects of TBWEP.

Conventional weed management approaches are inefficient and non-suitable for integration with smart agricultural machinery. Automatic identification and classification of weeds can play a vital role in weed management contributing to better crop yields. Intelligent and smart spot-spraying system's efficiency relies on the accuracy of the computer vision based detectors for autonomous weed control. In the present study, feasibility of deep learning based techniques (Alexnet, GoogLeNet, InceptionV3, Xception) were evaluated in weed identification from RGB images of bell pepper field. The models were trained with different values of epochs (10, 20,30), batch sizes (16, 32), and hyperparameters were tuned to get optimal performance. The overall accuracy of the selected models varied from 94.5 to 97.7%. Among the models, InceptionV3 exhibited superior performance at 30-epoch and 16-batch size with a 97.7% accuracy, 98.5% precision, and 97.8% recall. For this Inception3 model, the type 1 error was obtained as 1.4% and type II error was 0.9%. The effectiveness of the deep learning model presents a clear path towards integrating them with image-based herbicide applicators for precise weed management.

This paper aims to solve the lack of generalization of existing semantic segmentation models in the crop and weed segmentation domain. We compare two training mechanisms, classical and adversarial, to understand which scheme works best for a particular encoder-decoder model. We use simple U-Net, SegNet, and DeepLabv3+ with ResNet-50 backbone as segmentation networks. The models are trained with cross-entropy loss for classical and PatchGAN loss for adversarial training. By adopting the Conditional Generative Adversarial Network (CGAN) hierarchical settings, we penalize different Generators (G) using PatchGAN Discriminator (D) and L1 loss to generate segmentation output. The generalization is to exhibit fewer failures and perform comparably for growing plants with different data distributions. We utilize the images from four different stages of sugar beet. We divide the data so that the full-grown stage is used for training, whereas earlier stages are entirely dedicated to testing the model. We conclude that U-Net trained in adversarial settings is more robust to changes in the dataset. The adversarially trained U-Net reports 10% overall improvement in the results with mIOU scores of 0.34, 0.55, 0.75, and 0.85 for four different growth stages.

Artificial intelligence and machine learning have been increasingly applied for prediction in agricultural science. However, many models are typically black boxes, meaning we cannot explain what the models learned from the data and the reasons behind predictions. To address this issue, I introduce an emerging subdomain of artificial intelligence, explainable artificial intelligence (XAI), and associated toolkits, interpretable machine learning. This study demonstrates the usefulness of several methods by applying them to an openly available dataset. The dataset includes the no-tillage effect on crop yield relative to conventional tillage and soil, climate, and management variables. Data analysis discovered that no-tillage management can increase maize crop yield where yield in conventional tillage is <5000 kg/ha and the maximum temperature is higher than 32°. These methods are useful to answer (i) which variables are important for prediction in regression/classification, (ii) which variable interactions are important for prediction, (iii) how important variables and their interactions are associated with the response variable, (iv) what are the reasons underlying a predicted value for a certain instance, and (v) whether different machine learning algorithms offer the same answer to these questions. I argue that the goodness of model fit is overly evaluated with model performance measures in the current practice, while these questions are unanswered. XAI and interpretable machine learning can enhance trust and explainability in AI.

Artificial intelligence (AI) has advanced at an astounding rate and transformed numerous economic sectors. Nevertheless, a comprehensive understanding of how AI can improve the agri-food industry is lacking. In addition, there is a notable dearth of research on AI that investigates the influence of AI on agri-food resources and educates practitioners on the significance of knowledge-based and smart agriculture. We utilised bibliometric analysis to investigate the present state of the art and emerging trends in the relationship between AI and the agri-food industry. The research identified three distinct growth phases and the most prevalent AI strategies in the industry. In addition, we analysed key trends and offered researchers and practitioners insightful recommendations for future research. Using resource-based view (RBV) as the theoretical lens, this study established a framework emphasising the long-term effects of AI on various agri-food resources and proposed several research propositions. In addition, AI-related obstacles have been identified and categorised into four major categories. Lastly, the originality of the article lies in its numerous research suggestions and recommendations for advancing the AI field in the agri-food industry.