Corn is one of the most important feed and food crops in the world, and the purity and variety of its seeds are crucial to achieving high yields. Accurate corn variety identification is essential for optimizing post-harvest storage strategies, as different varieties exhibit distinct responses to pests, moisture, and temperature conditions. This paper proposes an efficient method for identifying corn varieties based on hyperspectral imaging technology, data fusion, and machine learning. A fusion dataset of five different corn seed varieties is proposed, utilizing both spectral and texture features to enhance the model's detection accuracy. The potential role of data preprocessing in improving detection accuracy is explored. Principal component analysis is used for feature selection and dimensionality reduction, reducing the computational load of the model. The PSX-Stacking algorithm is applied to establish a stacking ensemble learning model for the non-destructive identification of corn varieties, and the discrimination performance of models built using spectral data, texture feature data, and fused data was validated. The experimental results show that the PSX-Stacking model achieves 95.66 % accuracy when using single-source data. Based on fused data, the PSX-Stacking model achieves 99.33 % accuracy, and the stacked ensemble learning model consistently outperforms other models.

Egg price has the characteristics of non-stationary, non-linear, and high volatility, which is more difficult to predict accurately. In this paper, we comprehensively consider the multiple factors affecting egg prices and construct a sequence-to-sequence (Seq2seq) model to study the multi-step prediction method of egg prices. Seasonal-trend Decomposition Procedure Based on Loess (STL) is first used to decompose the historical egg price series into trend, seasonal, and residual terms to reduce the interference of sample noise on forecasting performance. Then, Principal Component Analysis (PCA) is used to analyze and downscale the multidimensional factors affecting egg prices, such as feed price, laying hen seedling price, culled chicken price, duck egg price, and consumer index, to eliminate the redundant information in the data. Finally, the above-processed data were introduced into the Seq2seq network for training to establish a multi-step prediction model for egg prices. The experimental results show that the STL-PCA-Seq2seq model proposed in this paper can broadly capture the long-term dependence information of the input series and model the complex nonlinear relationships among the multidimensional factors affecting egg prices with the lowest prediction errors compared to the Long Short-Term Memory (LSTM), Gated Recurrent Units (GRU), the Informer model, the Seq2seq model, and the STL-Seq2seq model. The method proposed in this paper can reach R2 of 0.9867, 0.9569, and 0.9106 at prediction steps 6, 12, and 18. With a prediction step size of 6, the RMSE is 0.131, MAE is 0.086, and MAPE is 0.813, respectively, which realizes the accurate prediction of egg price at any number of steps, and the results of the study provide a reference for the multi-step prediction of egg prices.

Precision agriculture technology is an essential means to improve the quality of crop production. The intelligent weeding robot is a new method of regulating agriculture. However, while balancing accuracy and model size, it is still a big challenge to solve the detection difficulties caused by the similar texture of soybean and weed leaves, mutual covering, light and shadow, and other factors. To address this, an improved You Only Look Once version 8 (YOLOv8n) is proposed in this paper. Firstly, the focusing diffusion pyramid network (FDPN) is used to optimize the structure and enhance the model detection ability. Then, a lightweight Contextual Continuous Convolution for FasterBlock with an efficient multiscale attention (C2f-F-EMA) structure is designed to simplify the model size and increase the model precision. At the same time, a lightweight detection head (LDH) is devised to cut the number of model parameters and make the model more lightweight. Finally, the Inner Complete Intersection over Union (Inner-CIoU) loss function is introduced to expedite the bounding box regression; it can improve the regression precision and training velocity. According to the article, the proposed algorithm performs well in weed classification, identification, and detection, and the precision, recall, and Mean Average Precision (mAP0.5) are 94.5 %, 92.6 %, and 97.2 %. The proposed model showed the best overall performance for other mainstream weed recognition algorithms, ensuring identification accuracy and lightweight performance.

Standardized striking movements are essential in badminton for enhancing player techniques and minimizing sports-related injuries. However, accurately detecting these movements against complex backgrounds while balancing precision and speed remains a significant challenge. To address this, we propose a novel model that synergizes You Only Look Once (YOLO) with the Hourglass Network (HGNet), called YOLO-HGNet, to enhance feature learning across multiple levels. By replacing traditional convolutional modules with Depth-Wise Convolution (DWConv), we achieve significant improvements in data processing efficiency. Additionally, our model incorporates a combination of self-attention and convolution mechanisms (ACmix) and FocalModulation to improve object localization and recognition accuracy in complex backgrounds. Our method leverages action vectors and machine learning techniques to accurately detect and classify six key badminton strokes: backhand push, backhand net shot, forehand clear, forehand push, forehand lift, and forehand net shot. Empirical evaluations demonstrate that our approach achieves a mean Average Precision (mAP) of 96.1% for detecting badminton player postures, outperforming existing advanced methods by at least 8.8%. Furthermore, our method achieves an average accuracy of 95.4% in classifying the six badminton strokes. These results underscore the superior capability of YOLO-HGNet for precise and efficient pose detection, recognition, and classification of badminton strokes, contributing significantly to advancements in sports science and athlete training methodologies.

含水率影响着花生的质量、储藏时长与出油率.本研究针对当前花生含水率测量效率低、有损检测、无法适应大规模检测等问题,探索基于高光谱成像技术的花生含水率无损快速检测方法.测量并建立了300份不同种类花生的高光谱原始图像及光谱数据集,并利用小波变换、多元散射校正(MSC)和一阶导数对数据进行预处理,结合PLS、XGBoost、BO-XGBoost算法建立花生含水量无损检测模型.通过实验对比得出,利用小波变换对原始光谱数据进行预处理后的光谱数据建立的BO-XGBoost模型最优,预测模型决定系数R2=0.953 9,均方根误差RMSE=0.806 5.实验表明,高光谱成像技术结合BO-XGBoost能够对花生含水率进行快速、准确、无损检测,且对其他农作物水分含量检测具有一定的借鉴意义.

Artificial intelligence is widely applied in sports science, especially in football, including match result prediction, and analysis of the performance of footballer. In team sports, the classification of footballer position can provide guidance for the selection of talents. However, there is still a lack of work in determining the position of the footballer, which is one of the leading problems for coaches in football. At present, it mainly relies on the preferences of footballers and arrangements of coaches. Therefore, the aim of this study is to predict the position of footballer using the dataset of football matches. In this study, we used web crawler to receive match dataset from the football analysis website whoscored. We tried to analyse the position of footballer and the factors affecting their positions. To achieve this aim, a two-stage application is followed. In the first stage, feature engineering was established to select important features. In the second stage, machine learning model was established. The result shows that compared with other models, the model based on GRU-attention has the highest accuracy (92.16%), the lowest loss (0.0144) and the fastest time. The AUC values of forwards, midfielders and defenders are 98%, 92% and 92%. The result verifies the effectiveness of the model.

It is of great practical significance to quickly, accurately, and effectively identify the effects of rice diseases on rice yield. This paper proposes a rice disease identification method based on an improved DenseNet network (DenseNet). This method uses DenseNet as the benchmark model and uses the channel attention mechanism squeeze-and-excitation to strengthen the favorable features, while suppressing the unfavorable features. Then, depth wise separable convolutions are introduced to replace some standard convolutions in the dense network to improve the parameter utilization and training speed. Using the AdaBound algorithm, combined with the adaptive optimization method, the parameter adjustment time reduces. In the experiments on five kinds of rice disease datasets, the average classification accuracy of the method in this paper is 99.4%, which is 13.8 percentage points higher than the original model. At the same time, it is compared with other existing recognition methods, such as ResNet, VGG, and Vision Transformer. The recognition accuracy of this method is higher, realizes the effective classification of rice disease images, and provides a new method for the development of crop disease identification technology and smart agriculture.

The evaluation of rice disease severity is a quantitative indicator for precise disease control, which is of great significance for ensuring rice yield. In the past, it was usually done manually, and the judgment of rice blast severity can be subjective and time-consuming. To address the above problems, this paper proposes a real-time rice blast disease segmentation method based on a feature fusion and attention mechanism: Deep Feature Fusion and Attention Network (abbreviated to DFFANet). To realize the extraction of the shallow and deep features of rice blast disease as complete as possible, a feature extraction (DCABlock) module and a feature fusion (FFM) module are designed; then, a lightweight attention module is further designed to guide the features learning, effectively fusing the extracted features at different scales, and use the above modules to build a DFFANet lightweight network model. This model is applied to rice blast spot segmentation and compared with other existing methods in this field. The experimental results show that the method proposed in this study has better anti-interference ability, achieving 96.15% MioU, a speed of 188 FPS, and the number of parameters is only 1.4 M, which can achieve a high detection speed with a small number of model parameters, and achieves an effective balance between segmentation accuracy and speed, thereby reducing the requirements for hardware equipment and realizing low-cost embedded development. It provides technical support for real-time rapid detection of rice diseases.

Egg freshness is an important economic index to measure egg quality, and it is also the main factor affecting egg sales. In this paper, aiming at the problem of small number of training and testing samples in current research, a sample collection device was set up, and 1173 pictures of egg samples with three different levels of freshness were collected, which greatly expanded the number of samples. On this basis, aiming at the problems of strong subjectivity and low accuracy of the obtained model when extracting features manually in the current research, the CBAM module is used in combination with the Inception module to construct a network model, and attention mechanism was introduced to assign adaptive weights to the collected multi-scale features, which further improved the accuracy of the network and the problem of network over-fitting, and establishes a high-precision egg freshness detection model. The test results showed that the average test accuracy of GoogLeNet-A reaches 94.05

垩白度是衡量优质大米品质的重要指标,随着农业检测自动化发展,利用机器视觉准确检测大米垩白度对大米生产加工具有重要意义.针对现有算法在分割垩白区域时存在抗干扰能力弱、稳定性差以及准确度低等问题,提出了一种基于图像显著性区域提取的垩白区域提取算法.利用大米垩白区域图像显著性的特点,对图像特征变化边缘进行提取,计算出边缘像素点个数以及边缘的总像素值,从而计算出边缘像素的平均值作为该区域的阈值.最后,利用计算得到的阈值对该区域进行分割,分割出整张图片的垩白区域,并计算出大米的垩白度.实验结果表明,该算法识别准确率为96.76％,相较于传统的OTSU算法检测准确率平均提高了26.87％,相较于改进的OTSU算法检测准确率平均提高了7.26％.

Aiming at the problems of high labor intensity and low efficiency in detecting dark spot eggs, a method of detecting dark spot eggs based on GoogLeNet model is proposed. This method uses Inception convolution module in GoogLeNet model to automatically extract dark spot eggs features and realize the detection. A device for collecting transparent images of eggs was set up in the experiment, and the sample collection experiments were designed to acquire samples. A total of 1200 dark spot eggs images and 8850 normal eggs images were obtained. Selecting 1200 samples of these two kinds for network modeling. The experimental results show that the detection accuracy of dark spotted eggs based on CNN GoogLeNet model is 98.19

Statistical models for predicting potential laying pattern were important for economically optimal breeding strategy of egg production in a poultry flock. The aim of this study was to establish an optimal model for describing egg production using room temperature, feed consumption, layer weight, and age during the production period. The following mathematical models were used PSO-LSSVM (Particle swarm optimization-Least squares support vector machines) and PCA (Principal component analysis). The daily recorded of egg production data from 19,666 laying-type hens was used. Hen-daily egg production was described using egg-laying rate during successive days after reaching sexual maturity (120 days of age) and daily recorded room temperature, feed consumption, layer weight, and age. Then present study used PCA to study the correlation between this data. Using the Pearson correlation coefficient of the five factors (maximum and minimum shed temperatures, layer weight, feed consumption, and age) and egg-laying rate, the present study weighted each factor according to its influence on the egg-laying rate. In addition, LSSVM was used to create a regression model of the weighted data, and PSO was to optimize parameters (penalty coefficient c and kernel parameter g) in the LSSVM. Our experimental results showed that the goodness-of-fit criteria value (MSE) was small, lower than that of existing prediction models. The PSO-LSSVM model was able to fit well egg-laying rate of the whole Hy-Line Brown laying-type hens' flock.

目前鸡蛋产量预测模型大多使用单一影响特征或者平均考虑各特征迸行建模,存在精度低、抗干扰能力差等缺点.针对上述问题,利用多层LSTM神经网络结合日龄、最高舍温、最低舍温、体质量、饲料消耗量5项特征建立高精度海兰褐蛋鸡产蛋率回归模型,并将得到的模型与传统的SVM模型和单层LSTM模型结果迸行对比.结果表明,本研究提出的利用多层LSTM模型预测鸡蛋产量均方误差更小,模型精度更高.

with the development of agricultural intelligence, it is of great significance to detect egg quality by machine vision and support vector machine in the field of food safety. In order to solve the problems of low efficiency and low accuracy of existing detection methods for egg crack detection, this paper proposes an egg crack detection and recognition method based on support vector machine and machine vision. The feature parameters of egg crack image are extracted by gray scale conversion, median filtering, linear sharpening, threshold segmentation and other means, and the support vector machine model is established, and the model is used to identify and detect eggs. The experimental results show that the model can distinguish intact eggs from cracked eggs, and the detection accuracy of cracked eggs is 98.75%.

专业培养方案是学生培养的规范和标准,课程体系是专业培养方案的主体之一,构建基于OBE理念,足以支撑毕业要求和培养目标的专业课程体系是为社会输送合格的工程教育人才,实现新工科建设,建成工程教育强国,形成中国特色世界一流工程教育体系的前提.文章基于浙江师范大学电子信息工程专业工程教育专业认证经验和实践,介绍了电子信息工程专业课程体系,各课程与认证标准的关系,各模块对学生能力的培养及与解决复杂工程问题能力培养的支撑关系.为同类专业修订符合工程教育专业认证要求的课程体系提供参考.

基于浙江师范大学电子信息工程专业(全国首家师范院校通过工程教育认证的专业)人才培养体系,课题组总结分析了专业培养目标、毕业要求和课程体系的制订优化和相互关系.课题组根据社会经济发展需求、学校定位确定了培养目标,确立了可衡量的毕业要求和足以支撑毕业要求的课程体系,同时建立了科学合理的培养目标和毕业要求修订机制和达成情况评价机制、毕业生跟踪反馈和社会评价机制、教学质量保障机制,通过跟踪反馈、评价和持续改进的思路实现了运行机制的良好运作和学生培养体系的不断优化.

基于图像步态识别因缺乏有效动态、时序特征,导致跨视角识别时准确率较低,而基于模型步态识别特征维度不足,容易造成步态识别平均准确率不高.故提出一种改进时空步态图(Improved Chrono-Gait Image,ICGI)及特征融合策略的解决方法,将时序信息与人体下肢关节间角度的规律变化相结合,突出步态运动时下肢的周期性变化.在引入时序信息的基础上,融合下肢关节点间动态特征,建立一个更加丰富、有效的特征集.结合最近邻算法(KNN)建立步态识别模型,在CASIA-B数据集上进行对比实验,证实所提方法能有效提高复杂环境下步态识别精度.

Power line communication (PLC) technology can make full use of the existing distribution network physical network for data transmission in smart grid, with low cost, flexibility, high coverage, network reliability advantages. In order to extend the communication distance of PLC network and improve the reliability of network, it is necessary to research the networking and deployment methods of PLC network and establish a reasonable optimization model for different environments and business requirements. Therefore, in order to solve the problem of PLC relay deployment under the business scenario of power communication network, this paper proposes an algorithm of PLC network relay station deployment for time delay optimization. The graph theory is used to describe and define the reliability of the network. In order to reduce time delay, the reliability and transmission power are defined as constraints. The mathematical model of relay station deployment is proposed. The improved genetic algorithm is designed and the relay station deployment algorithm based on the improved genetic algorithm is proposed. The performance of the algorithm is verified by simulation results.

蛋鸡产蛋率受生物、化学、物理以及人为等多方面因素影响,准确地预测蛋鸡产蛋率的变化趋势,建立蛋鸡的产蛋率预测模型对蛋鸡养殖具有重要的意义.将蛋鸡采食量、蛋鸡鸡龄、体质量、温度、光照时间以及是否服用营养素等6类影响因子进行处理,作为支持向量机(SVM)的输入数据,对蛋鸡的产蛋率进行预测,得到了一个稳定性好、适用范围广、预测结果准确的蛋鸡产蛋率模型,且预测结果符合蛋鸡的实际产蛋情况;同时为评估和分析SVM蛋鸡产蛋率预测模型的性能,以同样样本建立BP神经网络的预测模型,并用网络训练、测试用时、均方误差MSE以及相关系数r作为预测模型性能的评价指标.结果 表明,基于支持向量机的蛋鸡产蛋率预测模型精度和耗时均优于神经网络预测模型.