Applications of Machine Learning in Agriculture

Black soil plays an important role in maintaining a healthy ecosystem, promoting high-yield and efficient agricultural production, and conserving soil resources. In this paper, a typical black soil area of Keshan Farm in Qiqihar City, Heilongjiang Province, China, is used as a case study to investigate the black soil farmland productivity evaluation model. Based on the analysis of the composite index (CI) model, productivity index (PI) model and various machine learning models, the soil productivity evaluation method was improved and a prediction model was established. The results showed that the support vector machine regression model based on simulated annealing algorithm (SA-SVR), as well as the Gaussian process regression model (GPR), had obvious advantages in data preprocessing, feature selection, and model optimization compared to the modified composite index model (MCI), the modified productivity index model (MPI), and the coefficients of determination (R²) of their modelling, which were up to 0.70 and 0.71, respectively, and these machine learning prediction models can reflect the effects on maize cultivation and its yield through soil parameters even with small datasets, which can better capture the nonlinear relationship and improve the accuracy and stability of yield prediction, and is an effective method for guiding agricultural production as well as soil productivity evaluation. Full article

This study pioneered a non-destructive testing approach to evaluating the physicochemical properties of golden passion fruit by developing a platform to analyze the fruit’s electrical characteristics. By using dielectric properties, the method accurately predicted the soluble solids content (SSC), Acidity and pulp percentage (PP) in passion fruit. The investigation entailed measuring the relative dielectric constant (ε′) and dielectric loss factor (ε″) for 192 samples across a spectrum of 34 frequencies from 0.05 to 100 kHz. The analysis revealed that with increasing frequency and fruit maturity, both ε′ and ε″ showed a declining trend. Moreover, there was a discernible correlation between the fruit’s physicochemical indicators and dielectric properties. In refining the dataset, 12 outliers were removed using the Local Outlier Factor (LOF) algorithm. The study employed various advanced feature extraction techniques, including Recursive Feature Elimination with Cross-Validation (RFECV), Permutation Importance based on Random Forest Regression (PI-RF), Permutation Importance based on Linear Regression (PI-LR) and Genetic Algorithm (GA). All the variables and the selected variables after screening were used as inputs to build Extreme Gradient Boosting (XGBoost) and Categorical Boosting (Cat-Boost) models to predict the SSC, Acidity and PP in passion fruit. The results indicate that the PI-RF-XGBoost model demonstrated superior performance in predicting both the SSC (R² = 0.9240, RMSE = 0.2595) and the PP (R² = 0.9092, RMSE = 0.0014) of passion fruit. Meanwhile, the GA-CatBoost model exhibited the best performance in predicting Acidity (R² = 0.9471, RMSE = 0.1237). In addition, for the well-performing algorithms, the selected features are mainly concentrated within the frequency range of 0.05–6 kHz, which is consistent with the frequency range highly correlated with the dielectric properties and quality indicators. It is feasible to predict the quality indicators of fruit by detecting their low-frequency dielectric properties. This research offers significant insights and a valuable reference for non-destructive testing methods in assessing the quality of golden passion fruit. Full article

One of the main challenges agricultural greenhouses face is accurately predicting environmental conditions to ensure optimal crop growth. However, the current prediction methods have limitations in handling large volumes of dynamic and nonlinear temporal data, which makes it difficult to make accurate early predictions. This paper aims to forecast a greenhouse’s internal temperature up to one hour in advance using supervised learning tools like Extreme Gradient Boosting (XGBoost) and Recurrent Neural Networks combined with Long-Short Term Memory (LSTM-RNN). The study uses the many-to-one configuration, with a sequence of three input elements and one output element. Significant improvements in the R², RMSE, MAE, and MAPE metrics are observed by considering various combinations. In addition, Bayesian optimization is employed to find the best hyperparameters for each algorithm. The research uses a database of internal data such as temperature, humidity, and dew point and external data such as temperature, humidity, and solar radiation, splitting the data into the year’s four seasons and performing eight experiments according to the two algorithms and each season. The LSTM-RNN model produces the best results for the metrics in summer, achieving an R² = 0.9994, RMSE = 0.2698, MAE = 0.1449, and MAPE = 0.0041, meeting the acceptability criterion of ±2 °C hysteresis. Full article

For decades, humans have been confronted with numerous pest species, with the desert locust being one of the most damaging and having the greatest socio-economic impact. Trying to predict the occurrence of such pests is often complicated by the small number of records and observations in databases. This paper proposes a methodology based on a combination of classification and regression techniques to address not only the problem of locust sightings prediction, but also the number of locust individuals that may be expected. For this purpose, we apply different machine learning (ML) and related techniques, such as linear regression, Support Vector Machines, decision trees, random forests and neural networks. The considered ML algorithms are evaluated in three different scenarios in Western Africa, mainly Mauritania, and for the elaboration of the forecasting process, a number of meteorological variables obtained from the ERA5 reanalysis data are used as input variables for the classification–regression machines. The results obtained show good performance in terms of classification (appearance or not of desert locust), and acceptable regression results in terms of predicting the number of locusts, a harder problem due to the small number of samples available. We observed that the RF algorithm exhibited exceptional performance in the classification task (presence/absence) and achieved noteworthy results in regression (number of sightings), being the most effective machine learning algorithm among those used. It achieved classification results, in terms of F-score, around the value of 0.9 for the proposed Scenario 1. Full article

Image-based Arabian camel breed classification is an important task for various practical applications, such as breeding management, genetic improvement, conservation, and traceability. However, it is a challenging task due to the lack of standardized criteria and methods, the high similarity among breeds, and the limited availability of data and resources. In this paper, we propose an approach to tackle this challenge by using convolutional neural networks (CNNs) and transfer learning to classify images of six different Arabian camel breeds: Waddeh, Majaheem, Homor, Sofor, Shaele, and Shageh. To achieve this, we created, preprocessed, and annotated a novel dataset of 1073 camel images. We then pre-trained CNNs as feature extractors and fine-tuned them on our new dataset. We evaluated several popular CNN architectures with diverse characteristics such as InceptionV3, NASNetLarge, PNASNet-5-Large, MobileNetV3-Large, and EfficientNetV2 (small, medium, and large variants), and we found that NASNetLarge achieves the best test accuracy of 85.80% on our proposed dataset. Finally, we integrated the best-performing CNN architecture, NASNetLarge, into a mobile application for further validation and actual use in a real-world scenarios. Full article

The soluble solids content (SSC) is one of the important evaluation indicators for the internal quality of fresh grapes. However, the current non-destructive detection method based on hyperspectral imaging (HSI) relies on manual operation and is relatively cumbersome, making it difficult to achieve automatic detection in batches. Therefore, in this study, we aimed to conduct research on an improved non-destructive detection method for the SSC of bunch-harvested grapes. This study took the Shine-Muscat grape as the research object. Using Mask R-CNN to establish a grape image segmentation model based on deep learning (DL) applied to near-infrared hyperspectral images (400~1000 nm), 35 characteristic wavelengths were selected using Monte Carlo Uninformative Variable Elimination (MCUVE) to establish a prediction model for SSC. Based on the two abovementioned models, the improved non-destructive detection method for the SSC of bunch-harvested grapes was validated. The comprehensive evaluation index F₁ of the image segmentation model was 95.34%. The R_m² and RMSEM of the SSC prediction model were 0.8705 and 0.5696 Brix%, respectively, while the R_p² and RMSEP were 0.8755 and 0.9177 Brix%, respectively. The non-destructive detection speed of the improved method was 16.6 times that of the existing method. These results prove that the improved non-destructive detection method for the SSC of bunch-harvested grapes based on DL and HSI is feasible and efficient. Full article

This paper proposes a multi–convolutional neural network (CNN)-based system for the detection, tracking, and recognition of the emotions of dogs in surveillance videos. This system detects dogs in each frame of a video, tracks the dogs in the video, and recognizes the dogs’ emotions. The system uses a YOLOv3 model for dog detection. The dogs are tracked in real time with a deep association metric model (DeepDogTrack), which uses a Kalman filter combined with a CNN for processing. Thereafter, the dogs’ emotional behaviors are categorized into three types—angry (or aggressive), happy (or excited), and neutral (or general) behaviors—on the basis of manual judgments made by veterinary experts and custom dog breeders. The system extracts sub-images from videos of dogs, determines whether the images are sufficient to recognize the dogs’ emotions, and uses the long short-term deep features of dog memory networks model (LDFDMN) to identify the dog’s emotions. The dog detection experiments were conducted using two image datasets to verify the model’s effectiveness, and the detection accuracy rates were 97.59% and 94.62%, respectively. Detection errors occurred when the dog’s facial features were obscured, when the dog was of a special breed, when the dog’s body was covered, or when the dog region was incomplete. The dog-tracking experiments were conducted using three video datasets, each containing one or more dogs. The highest tracking accuracy rate (93.02%) was achieved when only one dog was in the video, and the highest tracking rate achieved for a video containing multiple dogs was 86.45%. Tracking errors occurred when the region covered by a dog’s body increased as the dog entered or left the screen, resulting in tracking loss. The dog emotion recognition experiments were conducted using two video datasets. The emotion recognition accuracy rates were 81.73% and 76.02%, respectively. Recognition errors occurred when the background of the image was removed, resulting in the dog region being unclear and the incorrect emotion being recognized. Of the three emotions, anger was the most prominently represented; therefore, the recognition rates for angry emotions were higher than those for happy or neutral emotions. Emotion recognition errors occurred when the dog’s movements were too subtle or too fast, the image was blurred, the shooting angle was suboptimal, or the video resolution was too low. Nevertheless, the current experiments revealed that the proposed system can correctly recognize the emotions of dogs in videos. The accuracy of the proposed system can be dramatically increased by using more images and videos for training the detection, tracking, and emotional recognition models. The system can then be applied in real-world situations to assist in the early identification of dogs that may exhibit aggressive behavior. Full article

Rose oil production is believed to be dependent on only a few genotypes of the famous rose Rosa damascena. The aim of this study was to develop a novel GC-MS fingerprint based on the need to expand the genetic resources of oil-bearing rose for industrial cultivation in the Taif region (Saudi Arabia). Gas chromatography-mass spectrometry (GC-MS) is a widely used analytical technique for determining the volatile composition of distilled rose oil from flower data. Because biosample availability, prohibitive costs, and ethical concerns limit observations in agricultural research, we aimed to enhance the quality of analysis by combining real observations with samples generated in silico. This study proposes a novel artificial intelligence model based on generative adversarial neural networks (GANs) to classify Taif rose cultivars using raw GC-MS data. We employed a variant of the GAN known as conditional stacked GANs (cSGANs) to predict Taif rose’s oil content and other latent characteristics without the need to conduct laboratory tests. A hierarchical stack of conditional GANs is used in this algorithm to generate images. A cluster model was developed based on the dataset provided, to quantify the diversity that should be implemented in the proposed model. The networks were trained using the cross-entropy and minimax loss functions. The accuracy of the proposed model was assessed by measuring losses as a function of the number of epochs. The results prove the ability of the proposed model to perfectly generate new real samples of different classes based on the GC-MS fingerprint. Full article

Recently, several plant pathogens have become more active due to temperature increases arising from climate change, which has caused damage to various crops. If climate change continues, it will likely be very difficult to maintain current crop production, and the problem of a shortage of expert manpower is also deepening. Fortunately, research on various early diagnosis systems based on deep learning is actively underway to solve these problems, but the problem of lack of diversity in some hard-to-collect disease samples remains. This imbalanced data increases the bias of machine learning models, causing overfitting problems. In this paper, we propose a data augmentation method based on an image-to-image translation model to solve the bias problem by supplementing these insufficient diseased leaf images. The proposed augmentation method performs translation between healthy and diseased leaf images and utilizes attention mechanisms to create images that reflect more evident disease textures. Through these improvements, we generated a more plausible diseased leaf image compared to existing methods and conducted an experiment to verify whether this data augmentation method could further improve the performance of a classification model for early diagnosis of plants. In the experiment, the PlantVillage dataset was used, and the extended dataset was built using the generated images and original images, and the performance of the classification models was evaluated through the test set. Full article