Seed germination rate is one of the important indicators in measuring seed quality and seed germination ability, and it is also an important basis for evaluating the growth potential and planting effect of seeds. In order to detect seed germination rates more efficiently and achieve automated detection, this study focuses on wild rice as the research subject. A novel method for detecting wild rice germination rates is introduced, leveraging the SGR-YOLO model through deep learning techniques. The SGR-YOLO model incorporates the convolutional block attention module (efficient channel attention (ECA)) in the Backbone, adopts the structure of bi-directional feature pyramid network (BiFPN) in the Neck part, adopts the generalized intersection over union (GIOU) function as the loss function in the Prediction part, and adopts the GIOU function as the loss function by setting the weighting coefficient to accelerate the detection of the seed germination rate. In the Prediction part, the GIOU function is used as the loss function to accelerate the learning of high-confidence targets by setting the weight coefficients to further improve the detection accuracy of seed germination rate. The results showed that the accuracy of the SGR-YOLO model for wild rice seed germination discrimination was 94% for the hydroponic box and 98.2% for the Petri dish. The errors of germination potential, germination index, and average germination days detected by SGR-YOLO using the manual statistics were 0.4%, 2.2, and 0.9 days, respectively, in the hydroponic box and 0.5%, 0.5, and 0.24 days, respectively, in the Petri dish. The above results showed that the SGR-YOLO model can realize the rapid detection of germination rate, germination potential, germination index, and average germination days of wild rice seeds, which can provide a reference for the rapid detection of crop seed germination rate.

Wild rice, a natural gene pool for rice germplasm innovation and variety improvement, holds immense value in rice breeding due to its disease-resistance genes. Traditional disease resistance identification in wild rice heavily relies on labor-intensive and subjective manual methods, posing significant challenges for large-scale identification. The fusion of unmanned aerial vehicles (UAVs) and deep learning is emerging as a novel trend in intelligent disease resistance identification. Detecting diseases in field conditions is critical in intelligent disease resistance identification. In pursuit of detecting bacterial blight in wild rice within natural field conditions, this study presents the Xoo-YOLO model, a modification of the YOLOv8 model tailored for this purpose. The Xoo-YOLO model incorporates the Large Selective Kernel Network (LSKNet) into its backbone network, allowing for more effective disease detection from the perspective of UAVs. This is achieved by dynamically adjusting its large spatial receptive field. Concurrently, the neck network receives enhancements by integrating the GSConv hybrid convolution module. This addition serves to reduce both the amount of calculation and parameters. To tackle the issue of disease appearing elongated and rotated when viewed from a UAV perspective, we incorporated a rotational angle (theta dimension) into the head layer's output. This enhancement enables precise detection of bacterial blight in any direction in wild rice. The experimental results highlight the effectiveness of our proposed Xoo-YOLO model, boasting a remarkable mean average precision (mAP) of 94.95%. This outperforms other models, underscoring its superiority. Our model strikes a harmonious balance between accuracy and speed in disease detection. It is a technical cornerstone, facilitating the intelligent identification of disease resistance in wild rice on a large scale.

Cotton, a vital textile raw material, is intricately linked to people’s livelihoods. Throughout the cotton cultivation process, various diseases threaten cotton crops, significantly impacting both cotton quality and yield. Deep learning has emerged as a crucial tool for detecting these diseases. However, deep learning models with high accuracy often come with redundant parameters, making them challenging to deploy on resource-constrained devices. Existing detection models struggle to strike the right balance between accuracy and speed, limiting their utility in this context. This study introduces the CDDLite-YOLO model, an innovation based on the YOLOv8 model, designed for detecting cotton diseases in natural field conditions. The C2f-Faster module replaces the Bottleneck structure in the C2f module within the backbone network, using partial convolution. The neck network adopts Slim-neck structure by replacing the C2f module with the GSConv and VoVGSCSP modules, based on GSConv. In the head, we introduce the MPDIoU loss function, addressing limitations in existing loss functions. Additionally, we designed the PCDetect detection head, integrating the PCD module and replacing some CBS modules with PCDetect. Our experimental results demonstrate the effectiveness of the CDDLite-YOLO model, achieving a remarkable mean average precision (mAP) of 90.6%. With a mere 1.8M parameters, 3.6G FLOPS, and a rapid detection speed of 222.22 FPS, it outperforms other models, showcasing its superiority. It successfully strikes a harmonious balance between detection speed, accuracy, and model size, positioning it as a promising candidate for deployment on an embedded GPU chip without sacrificing performance. Our model serves as a pivotal technical advancement, facilitating timely cotton disease detection and providing valuable insights for the design of detection models for agricultural inspection robots and other resource-constrained agricultural devices.

To find an economical solution to infer the depth of the surrounding environment of unmanned agricultural vehicles (UAV), a lightweight depth estimation model called MonoDA based on a convolutional neural network is proposed. A series of sequential frames from monocular videos are used to train the model. The model is composed of two subnetworks—the depth estimation subnetwork and the pose estimation subnetwork. The former is a modified version of U-Net that reduces the number of bridges, while the latter takes EfficientNet-B0 as its backbone network to extract the features of sequential frames and predict the pose transformation relations between the frames. The self-supervised strategy is adopted during the training, which means the depth information labels of frames are not needed. Instead, the adjacent frames in the image sequence and the reprojection relation of the pose are used to train the model. Subnetworks’ outputs (depth map and pose relation) are used to reconstruct the input frame, then a self-supervised loss between the reconstructed input and the original input is calculated. Finally, the loss is employed to update the parameters of the two subnetworks through the backward pass. Several experiments are conducted to evaluate the model’s performance, and the results show that MonoDA has competitive accuracy over the KITTI raw dataset as well as our vineyard dataset. Besides, our method also possessed the advantage of non-sensitivity to color. On the computing platform of our UAV’s environment perceptual system NVIDIA JETSON TX2, the model could run at 18.92 FPS. To sum up, our approach provides an economical solution for depth estimation by using monocular cameras, which achieves a good trade-off between accuracy and speed and can be used as a novel auxiliary depth detection paradigm for UAVs.

The grain output of China increased continuously for 11 years from 2004 to 2014, effectively ensuring security of the country’s food supply. At the same time, rapid advancement of urbanization and industrialization generated marked changes in the food consumption patterns of residents, and supply-demand structural problems, such as the variety, quality and region of grains, became increasingly prominent. Currently, national food security goes beyond the guarantee of total grain demand. Therefore, both the production and consumption systems must be evaluated to produce a scientific measurement of food security. From the perspective of food production-consumption coordination and matching, this article sets up a multidimensional coupling assessment index system and model, and carries out assessment of the food security level and the warning status of China between 1995 and 2012. Results show that the level of quantity coordination remained high and the level of variety coordination fluctuated before 2006 and continued decreasing afterwards. The regional coordination level largely continued to decline during the whole research period. The level of coordination of Chinese food production-consumption was high overall. The warning situation existed only for four years, but it started to decrease continuously since 2007 because of aggravated structural and regional disharmony. The state of coordination at the early stage primarily reflected variation of food production, but that of the later stage greatly reflected the problems caused by food consumption. In the future, food security of China can be ensured overall, but regional and variety coordination problems may worsen further, which may exert a potentially negative influence on agricultural trade, industrial security and market stability, thereby leading to decrease in food security at the regional and household levels. In the future, this problem can be addressed by ensuring the self-sufficiency of core varieties of grains, building interest coordination mechanisms in the production and sale regions, forming a modern food and nutrition development pattern, and making additional efforts to save grains.

Verticillium wilt (VW) represents the most formidable challenge in cotton cultivation, critically impairing both fiber yield and quality. Conventional resistance assessment techniques, which are largely reliant on subjective manual evaluation, fail to meet the demands for precision and scalability required for advanced genetic research. This study introduces a robust evaluation framework utilizing feature selection and optimization algorithms to enhance the accuracy and efficiency of the severity assessment of cotton VW. We conducted comprehensive time-series UAV hyperspectral imaging (400 to 995 nm) on the cotton canopy in a field environment on different days after sowing (DAS). After preprocessing the hyperspectral data to extract wavelet coefficients and vegetation indices, various feature selection methods were implemented to select sensitive spectral features for cotton VW. By leveraging these selected features, we developed machine learning models to assess the severity of cotton VW at the canopy scale. Model validation revealed that the performance of the assessment models responded dynamically as VW progressed and achieved the highest R² of 0.5807 at DAS 80, with an RMSE of 6.0887. Optimization algorithms made a marked improvement for SVM in severity assessment using all observation data, with R² increasing from 0.6986 to 0.9007. This study demonstrates the potential of feature selection and machine learning methods based on hyperspectral data in enhancing VW management, promising advancements in high-throughput automated disease assessment, and supporting sustainable agricultural practices.