Measurement of plant nitrogen (N), phosphorus (P), and potassium (K) levels are important for determining precise fertilization management approaches for crops cultivated in greenhouses. To accurately, rapidly, stably, and nondestructively measure the NPK levels in tomato plants, a nondestructive determination method based on multispectral three-dimensional (3D) imaging was proposed. Multiview RGB-D images and multispectral images were synchronously collected, and the plant multispectral reflectance was registered to the depth coordinates according to Fourier transform principles. Based on the Kinect sensor pose estimation and self-calibration, the unified transformation of the multiview point cloud coordinate system was realized. Finally, the iterative closest point (ICP) algorithm was used for the precise registration of multiview point clouds and the reconstruction of plant multispectral 3D point cloud models. Using the normalized grayscale similarity coefficient, the degree of spectral overlap, and the Hausdorff distance set, the accuracy of the reconstructed multispectral 3D point clouds was quantitatively evaluated, the average value was 0.9116, 0.9343 and 0.41 cm, respectively. The results indicated that the multispectral reflectance could be registered to the Kinect depth coordinates accurately based on the Fourier transform principles, the reconstruction accuracy of the multispectral 3D point cloud model met the model reconstruction needs of tomato plants. Using back-propagation artificial neural network (BPANN), support vector machine regression (SVMR), and gaussian process regression (GPR) methods, determination models for the NPK contents in tomato plants based on the reflectance characteristics of plant multispectral 3D point cloud models were separately constructed. The relative error (RE) of the N content by BPANN, SVMR and GPR prediction models were 2.27%, 7.46% and 4.03%, respectively. The RE of the P content by BPANN, SVMR and GPR prediction models were 3.32%, 8.92% and 8.41%, respectively. The RE of the K content by BPANN, SVMR and GPR prediction models were 3.27%, 5.73% and 3.32%, respectively. These models provided highly efficient and accurate measurements of the NPK contents in tomato plants. The NPK contents determination performance of these models were more stable than those of single-view models.

Soil acts as a foundation for human survival and social development and soil quality has a great effect on the growth of agricultural products. Visible/near-infrared spectroscopy has been acknowledged as a rapid and non-destructive method for predicting soil properties, and multi-task learning is a preferable approach to analyze the correlation between the spectroscopy data and soil properties. However, current multi-task learning models with the soft parameter sharing structure extremely rely on the task relatedness. To tackle this limitation, we proposed PS-MTL-LUCAS, a multi-task learning with a partially shared structure in this study. An additional shared layer was utilized to learn the general informative representations and interact with each task-specific layer. The partially shared structure ensured the maximum information flow between layers, thereby boosting the prediction performance. Also, the SHapley Addictive exPlanations (SHAP) algorithm was adopted to extract the feature wavelengths of each soil property. PS-MTL-LUCAS was validated on the LUCAS topsoil dataset (2009), and the experimental result suggested that PS-MTL-LUCAS dominated state-of-the-art models by achieving the determination of coefficient at 0.945, 0.936, 0.413, 0.624, 0.837, 0.952, and 0.956 for pH, N, P, K, CEC, OC, and CaCO₃, respectively. In summary, this study highlighted the use of the soil spectroscopy and multi-task learning techniques in the soil property prediction task and provided a very promising approach for soil studies.

Rice is a staple food for over half of the global population, but it faces significant yield losses: up to 52% due to leaf blast disease and brown spot diseases, respectively. This study aimed at proposing a hybrid architecture, namely ResViT-Rice, by taking advantage of both CNN and transformer for accurate detection of leaf blast and brown spot diseases. We employed ResNet as the backbone network to establish a detection model and introduced the encoder component from the transformer architecture. The convolutional block attention module was also integrated to ResViT-Rice to further enhance the feature-extraction ability. We processed 1648 training and 104 testing images for two diseases and the healthy class. To verify the effectiveness of the proposed ResViT-Rice, we conducted comparative evaluation with popular deep learning models. The experimental result suggested that ResViT-Rice achieved promising results in the rice disease-detection task, with the highest accuracy reaching 0.9904. The corresponding precision, recall, and F1-score were all over 0.96, with an AUC of up to 0.9987, and the corresponding loss rate was 0.0042. In conclusion, the proposed ResViT-Rice can better extract features of different rice diseases, thereby providing a more accurate and robust classification output.

The suitability of airflow velocity in airborne spraying operations in orchards is mostly evaluated on the basis of inlet and outlet based on the airflow velocity at the canopy. However, the airflow velocity required to penetrate into the inner layer of the canopy, which is prone to pests and diseases, is still unclear due to variation in the geometry of the plant canopies. In this study, pear trees were selected as an example to explore the variations in the law of airflow attenuation in the inner canopy. Furthermore, we examine mist droplet formation in the inner canopy to determine a suitable inner canopy airflow end velocity (ICAEV) for air-assisted application. We also conducted a field validation test. The results showed that the majority of airflow velocity loss occurred in the middle and outer part of the canopy; rapid decline of airflow occurred in the 0–0.3 m section, whereas the slow decline of airflow occurred in the 0.3–0.8 m section. When the ICAEV is in the range of 2.70–3.18 m/s, the spraying effect is better. The droplet deposition variation coefficient was 42.25% compared with 51.25% in the conventional airflow delivery mode. Additionally, the droplet drift was reduced by 12.59 μg/cm². The results of this study can identify a suitable ICAEV for air-assisted spraying in orchards.

Seed-fertilizer drill machines with fluted rollers generally utilize a calibrated relationship between the fluted-roller rotation speed and mass flow rate to control the application rate. However, this relationship model gradually deteriorates with operating time and is easily affected by working conditions. To maintain the initial operating accuracy, a self-calibrating system for adjusting the control model parameters was designed. It utilizes the application-rate information and fluted-rollers rotation speed data during the operation process to dynamically establish the calibrated relationship. During the field tests performed 18 times, the self-calibrating system could dynamically re-calibrate the relationship model. After the model parameters were updated, the application rate control system maintained a high accuracy level, though the relative deviation of control parameters was relatively higher. The average relative deviations between actual and theoretical cumulative application rates were 0.97% and 1.22% for seeding and fertilizing, respectively, and the standard deviations were 0.69% and 0.62%. Correspondingly, the average relative deviations of control parameters were 7.29% and 7.86%, and the standard deviations were 1.16% and 3.06%. These results indicate that the proposed method with closed-looped control can maintain high-quality control performance and improve the productivity of the drill machine, which will benefit the agricultural production.

Background: The characteristics of light source have an important influence on the measurement performance of canopy reflectance spectrometer. The size of the effective irradiation area and the uniformity of the light intensity distribution in the irradiation area determine the ability of the spectrometer to express the group characteristics of the measured objects. Methods: In this paper, an evaluation method was proposed to theoretically analyze the influence of the light intensity distribution characteristics of the light source irradiation area on the measurement results. The light intensity distribution feature vector and the reflectance feature vector of the measured object were constructed to design reflectance difference coefficient, which could effectively evaluate the measurement performance of the canopy reflectance spectrometer. By using self-design light intensity distribution test system and GreenSeeker RT100, the evaluation method was applied to evaluate the measurement results. Results: The evaluation results showed that the vegetation indices based on the arithmetic average reflectance of the measured object could be obtained theoretically only when the light intensity distribution of the light source detected by the spectrometer was uniform, which could fully express the group characteristics of the object. When the light intensity distribution of the active light source was not uniform, the measure value was difficult to fully express the group characteristics of the object. And the measured object reflectance was merely the weighted average value based on the light intensity distribution characteristics. Conclusions: According to the research results of this paper, sunlight is the most ideal detection light source. If the passive light source spectrometer can improve the measurement method to adapt to the change of sunlight intensity, its measurement performance will be better than any active-light spectrometer.