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

Spraying is currently one of the main methods of pesticide application worldwide. It converts the pesticide solution into fine droplets through a sprayer, which then deposit onto target plants. Therefore, in the process of pesticide application, improving the effectiveness of spraying while minimizing or preventing crop damage has become a key issue. Combining the advantages of electrostatic spraying technology with the characteristics of ground boom sprayers, a contact-type electrostatic boom spraying system based on a rod–plate electrode structure was designed and tested on a self-propelled boom sprayer. The charging chamber was designed based on the characteristics of the rod–plate electrode and theoretical analysis. The reliability of the device was verified through COMSOL numerical simulations, charge-to-mass ratio, droplet size, and droplet size spectrum measurements, and a droplet size prediction model was established. The deposition characteristics in soybean fields were analyzed using the Box–Behnken experimental design method. The results showed that the rod–plate electrode structure demonstrated its superiority with a maximum spatial electric field of 2.31 × 10⁶ V/m. When the spray pressure was 0.3 MPa and the charging voltage was 8 kV, the droplet size decreased by 26.6%, and the charge-to-mass ratio reached 2.88 mC/kg. Field experiments showed that when the charging voltage was 8 kV, the spray pressure was 0.3 MPa, the traveling speed was 7 km/h, and the number of deposited droplets was 8517. This study provides some basis for the application of electrostatic spraying technology in large-scale field operations.

Industrial hemp has significant utilization value, and China is the largest producer of industrial hemp in the world, primarily growing hemp for fiber. Heilongjiang Province is the largest area of hemp fiber cultivation in China. In response to issues such as inconsistent operating standards, low efficiency, and poor harvesting quality of industrial hemp harvesters in China, this study integrates the mechanical properties of the “Hanma 5” hemp stalk and applies cutting platform design theory to analyze and optimize the key components of existing hemp harvesters, aiming to obtain optimal operational parameters. First, by analyzing the motion laws of the cutting blade and the regression equations of shear power consumption and shear force, the relationship between cutting speed and time is established, and the cutting and conveying parameter ranges are derived, providing the basis for subsequent simulation analysis and field validation. Next, dynamic simulation analysis of the key components of the domestic 4GM-2.2 industrial hemp harvester for fiber is conducted using ADAMS and Workbench. The conveyor speed values and corresponding chain drive combinations under different conveyor speed ratios are obtained. Field experiments validated the optimal combination of key operational parameters for the industrial hemp harvester as follows: forward speed of 2.1 m/s, cutting speed of 2.5 m/s, conveyor speed ratio of 2.2, coefficient of variation for the laying angle of 6.88%, coefficient of variation for the laying thickness of 4.11%, and cutting height of 10.4 cm. Under the optimal operational parameter combination, the re-cut rate was 8.4%, with no missed cutting observed. This paper provides technical references for exploring the optimal operational parameters of industrial hemp harvesters for fiber to achieve high-quality harvesting operations.

To optimize the flow stability and improve application accuracy of the PWM intermittent variable-rate spraying system, which suffers from insufficient flow stability and response delays during changes in travel speed, this study proposes an intelligent control method based on an improved Adaptive Neural Fuzzy Inference System (ANFIS). Flow characteristic data of the solenoid valve were collected under four pressure conditions (0.2–0.5 MPa), drive frequencies (5–20 Hz), and duty cycles (10–90%) using an indoor test system. An ANFIS controller architecture was constructed with target flow rate and actual travel speed as input variables and PWM frequency-duty cycle combinations as output variables. This controller enhances the traditional single-output mode of ANFIS by achieving multi-output collaborative optimization through shared premise parameters, thereby strengthening the system’s nonlinear modeling and control capabilities. To validate the system’s practical performance, a field simulation test platform based on a spraying robot was constructed. By analyzing preset prescription map information, the system achieved precise variable-rate spraying operations during movement. Test results demonstrate that the steady-state error remains within 5.03% under various speed-varying conditions. This research provides a high-precision intelligent control solution for variable-rate spraying systems, holding significant implications for reducing pesticide application rates and advancing precision agriculture.

Dairy cow body temperature is a key physiological indicator that reflects metabolic level, immune status, and environmental stress responses, and it has been widely used for early disease recognition. Infrared thermography (IRT), as a non-contact imaging technique capable of remotely acquiring the surface radiation temperature distribution of animals, is regarded as a powerful alternative to traditional temperature measurement methods. Under practical cowshed conditions, IRT images of dairy cows are easily affected by complex background interference and generally suffer from low resolution, poor contrast, indistinct boundaries, weak structural perception, and insufficient texture information, which lead to significant degradation in target detection and temperature extraction performance. To address these issues, a lightweight detection model named YOLOv11n-CGSD is proposed for dairy cow IRT images, aiming to improve the accuracy and robustness of region of interest (ROI) detection and body temperature extraction under complex background conditions. At the architectural level, a C3Ghost lightweight module based on the Ghost concept is first constructed to reduce redundant feature extraction while lowering computational cost and enhancing the network capability for preserving fine-grained features during feature propagation. Subsequently, a space-to-depth convolution module is introduced to perform spatial rearrangement of feature maps and achieve channel compression via non-strided convolution, thereby improving the sensitivity of the model to local temperature variations and structural details. Finally, a dynamic sampling mechanism is embedded in the neck of the network, where the upsampling and scale alignment processes are adaptively driven by feature content, enhancing the model response to boundary temperature changes and weak-texture regions. Experimental results indicate that the YOLOv11n-CGSD model can effectively shift attention from irrelevant background regions to ROI contour boundaries and increase attention coverage within the ROI. Under complex IRT conditions, the model achieves P, R, and mAP50 values of 89.11%, 86.80%, and 91.94%, which represent improvements of 3.11%, 5.14%, and 4.08%, respectively, compared with the baseline model. Using T_max as the temperature extraction parameter, the maximum error (Max. Error) and mean error (MAE. Error) in the lower udder region are reduced by 33.3% and 25.7%, respectively, while in the around the anus region, the Max. Error and MAE. Error are reduced by 87.5% and 95.0%, respectively. These findings demonstrate that, under complex backgrounds and low-quality IRT imaging conditions, the proposed model achieves lightweight and high-performance detection for both lower udder (LU) and around the anus (AA) regions and provides a methodological reference and technical support for non-contact body temperature measurement of dairy cows in practical cowshed production environments.

The near-nozzle liquid sheet is a precursor to droplet formation, yet the quantitative relationship between its visual signatures and downstream droplet volume median diameter ($D_{\mathrm{v}50}$) remains insufficiently characterized. This study established a replicate-matched quantitative link between liquid-sheet visual signatures and downstream $D_{\mathrm{v}50}$ in flat-fan sprays. Three flat-fan nozzles were evaluated at 0.2, 0.3, and 0.4 MPa across 20 complete replicate blocks, yielding 180 replicate-matched records. Each record combined visual descriptors averaged from 500 high-speed backlit frames with one laser-measured $D_{\mathrm{v}50}$ value obtained under the same stabilized nozzle–pressure condition. YOLO-SpraySheet delineated the liquid sheet, from which the liquid-film pixel ratio and four gray-level co-occurrence matrix features were extracted. As pressure increased, the liquid-film pixel ratio decreased, Contrast increased, and angular second moment, Correlation, Homogeneity, and $D_{\mathrm{v}50}$ decreased consistently across all nozzles. The segmentation model achieved a mask mAP50-95 of 84.2%. Nozzle-specific models produced predictive $Q^2$ values of 0.695–0.828 and prediction root mean square errors of 6.04–9.92 μm. These results demonstrate that liquid-sheet morphology and texture contain quantifiable precursor information for nozzle-specific $D_{\mathrm{v}50}$ estimation under controlled conditions.