Unmanned aerial vehicles (UAVs) equipped with dual-band crop-growth sensors can achieve high-throughput acquisition of crop-growth information. However, the downwash airflow field of the UAV disturbs the crop canopy during sensor measurements. To resolve this issue, we used computational fluid dynamics (CFD), numerical simulation, and three-dimensional airflow field testers to study the UAV-borne multispectral-sensor method for monitoring crop growth. The results show that when the flying height of the UAV is 1 m from the crop canopy, the generated airflow field on the surface of the crop canopy is elliptical, with a long semiaxis length of about 0.45 m and a short semiaxis of about 0.4 m. The flow-field distribution results, combined with the sensor’s field of view, indicated that the support length of the UAV-borne multispectral sensor should be 0.6 m. Wheat test results showed that the ratio vegetation index (RVI) output of the UAV-borne spectral sensor had a linear fit coefficient of determination (R²) of 0.81, and a root mean square error (RMSE) of 0.38 compared with the ASD Fieldspec2 spectrometer. Our method improves the accuracy and stability of measurement results of the UAV-borne dual-band crop-growth sensor. Rice test results showed that the RVI value measured by the UAV-borne multispectral sensor had good linearity with leaf nitrogen accumulation (LNA), leaf area index (LAI), and leaf dry weight (LDW); R² was 0.62, 0.76, and 0.60, and RMSE was 2.28, 1.03, and 10.73, respectively. Our monitoring method could be well-applied to UAV-borne dual-band crop growth sensors.

Portable spectrometers can extract crop growth information in a non-destructive, rapid, and efficient way. However, most existing instruments can only be used to collect spectral reflectance and calculate vegetation indices. They are costly and lack crop spectral monitoring models and diagnosis models; therefore, crop growth information cannot be directly output from these instruments. In this paper, an active light source apparatus for crop growth monitoring and diagnosis (CGMD) was developed based on the canopy shape characteristics and spectral monitoring mechanisms of row-cultivated crops. A modulated light source was used in this apparatus to effectively eliminate the interference of ambient light. At the same time, a high-pass filter was used in the apparatus to extract the canopy reflectance spectral information, which improves the signa-noise ratio. Crop canopy vegetation indexes—the normalized difference vegetation index (NDVI) and the ratio vegetation index (RVI)—and crop growth parameters—leaf area index (LAI), leaf dry weight (LDW), leaf nitrogen accumulation (LNA), and leaf nitrogen content (LNC)—were obtained in real time. The ambient light influence test and the comparison test of different test heights of the CGMD apparatus show that it has high test stability and can effectively overcome ambient light interference within the standard test height range. The rice and wheat experimental results demonstrated that the test results of the CGMD apparatus and the commercial Analytical Spectral Devices field spectrometer (Malvern Panalytical, Malvern, Worcs, UK) were consistent. Comparing the vegetation indexes and agronomic parameters obtained from the chemical analysis, the CGMD apparatus has a good predictive ability for LAI, LDW, LNC, and LNA.

Introduction: Recognizing wheat ears plays a crucial role in predicting wheat yield. Employing deep learning methods for wheat ears identification is the mainstream method in current research and applications. However, such methods still face challenges, such as high computational parameter volume, large model weights, and slow processing speeds, making it difficult to apply them for real-time identification tasks on limited hardware resources in the wheat field. Therefore, exploring lightweight wheat ears detection methods for real-time recognition holds significant importance.Methods: This study proposes a lightweight method for detecting and counting wheat ears based on YOLOv5s. It utilizes the ShuffleNetV2 lightweight convolutional neural network to optimize the YOLOv5s model by reducing the number of parameters and simplifying the complexity of the calculation processes. In addition, a lightweight upsampling operator content-aware reassembly of features is introduced in the feature pyramid structure to eliminate the impact of the lightweight process on the model detection performance. This approach aims to improve the spatial resolution of the feature images, enhance the effectiveness of the perceptual field, and reduce information loss. Finally, by introducing the dynamic target detection head, the shape of the detection head and the feature extraction strategy can be dynamically adjusted, and the detection accuracy can be improved when encountering wheat ears with large-scale changes, diverse shapes, or significant orientation variations.Results and discussion: This study uses the global wheat head detection dataset and incorporates the local experimental dataset to improve the robustness and generalization of the proposed model. The weight, FLOPs and mAP of this model are 2.9 MB, 2.5 * 10⁹ and 94.8%, respectively. The linear fitting determination coefficients R² for the model test result and actual value of global wheat head detection dataset and local experimental Site are 0.94 and 0.97, respectively. The improved lightweight model can better meet the requirements of precision wheat ears counting and play an important role in embedded systems, mobile devices, or other hardware systems with limited computing resources.

The high-flux acquisition of crop growth information can be realized using field monitoring robotic platforms. However, most of the existing agricultural monitoring robots have been converted from expensive commercial platforms, and they thus have a hard time adapting to the farmland working environment, let alone satisfying the basic requirements of sensor testing. To address these problems, a wheeled crop-growth-monitoring robot that features the accurate, nondestructive, and efficient acquisition of crop growth information was developed based on the cultivation characteristics of wheat, the obstacle characteristics of the wheat field, and the monitoring mechanism of spectral sensors. By analyzing the phenotypic structural change characteristics and the requirements for the row spacing of different wheat varieties throughout the growth period, a four-wheel mobile chassis was designed with an adjustable wheel track and a high-clearance body structure that can effectively eliminate the risk of the robot destroying the wheat during operation. Moreover, considering the requirements for wheeled robots to overcome obstacles in field operations, a three-dimensional (3D) model of the robot was created in Pro/E. Models of obstacles in the field (e.g., pits and bumps) were created in Adams to simulate the operational stability of the robot. The simulation results showed that the mass center displacement of the robot was smaller than 0.2 cm on flat pavement and the maximum mass center displacement was 1.78 cm during obstacle crossing (10 cm deep pits and 10 cm high bumps). The field test showed that the robot equipped with active-light-source crop growth sensors achieved stable, real-time, nondestructive, and accurate acquisition of the canopy vegetation parameters—NDVI (normalized difference vegetation index) and RVI (ratio vegetation index)—and the wheat growth parameters—LAI (leaf area index), LDW (leaf dry weight), LNA (leaf nitrogen accumulation), and LNC (leaf nitrogen content).

Accurate identification of rice ears is crucial for assessing rice yield. Present research mainly relies on single-scale image data for rice ears detection and counting. However, these approaches are susceptible to misdetection and omission due to the intricate environmental conditions in fields. The combination of multi-source images can better overcome the limitations of single-scale images. In this study, based on the YOLOv5s target detection algorithm, a method for rice ears detection and counting applicable to multi-source images is proposed by integrating image data collected by cell phones and UAVs during the rice heading and maturity periods. The proposed method introduces Attention-based Intrascale Feature Interaction (AIFI) to reconstruct the backbone feature extraction network, optimizing feature expression interaction and enhancing handling of the model of advanced semantic information. Additionally, Simplify Optimal Transport Assignment (SimOTA) is employed to achieve a more refined label assignment strategy, thereby optimizing detection performance of the model and addressing difficulties in detecting multiple targets in high-density rice ears environments. Finally, Channel-wise Knowledge Distillation for Dense Prediction (CWD) is utilized to enhance the performance of the model in dense prediction tasks by transferring knowledge between different channels. The experimental results demonstrated good performance of the model on datasets comprising rice at the heading and maturity stages, achieving Precision, Recall, and mAP values of 93%, 85.3%, and 90.3%, respectively. The coefficients of determination (R $^{2}$ ) for the linear fit between test results and the actual statistical results of the model were 0.91, 0.91, 0.90, and 0.88, respectively. The proposed model performs well in the mixed dataset and can be utilized more effectively for accurate identification and counting of rice ears.