In agriculture, seed sorting is critical for production and marketing purposes. Low‐quality seeds can cause poor plant growth and lead to problems such as disease and low yields. This study uses machine vision and machine learning to develop a rapid detection and classification method for maize (Zea mays L.) seeds based on variety purity. A computer vision system was designed to recognize five varieties of maize seeds. Halogen lamps were applied for illumination and a high‐resolution RGB camera was used to acquire images of 8,080 maize seeds in the laboratory. An image processing algorithm was proposed to extract 16 important features (12 dimensional and 4 of shape) from the maize seed images, and a user‐friendly interface was developed using a MATLAB graphical user interface (GUI). Multilayer perceptron (MLP), decision tree (DT), linear discrimination (LDA), Naive Bayes (NB), support vector machine (SVM), k‐nearest neighbors (KNN), and AdaBoost algorithm were used to develop the varietal classification model. The optimal model parameters were obtained with 10‐fold cross‐validation, and the performance metrics were compared. The names of the maize varieties were marked in the GUI. The overall classification accuracy was determined as 96.26, 94.95, 95.97, 93.97, 96.46, 95.59, and 95.31% for MLP, DT, LDA, NB, SVM, KNN, and AdaBoost, respectively. The SVM classification model obtained the highest accuracy for BaoQiu, ShanCu, XinNuo, LiaoGe, and KouXian varieties, which reached 93.07, 98.95, 96.15, 89.65, and 99.22%, respectively. The classification results satisfy the needs of producers and consumers.

Maize is one of the essential crops for food supply. Accurate sorting of seeds is critical for cultivation and marketing purposes, while the traditional methods of variety identification are time-consuming, inefficient, and easily damaged. This study proposes a rapid classification method for maize seeds using a combination of machine vision and deep learning. 8080 maize seeds of five varieties were collected, and then the sample images were classified into training and validation sets in the proportion of 8:2, and the data were enhanced. The proposed improved network architecture, namely P-ResNet, was fine-tuned for transfer learning to recognize and categorize maize seeds, and then it compares the performance of the models. The results show that the overall classification accuracy was determined as 97.91, 96.44, 99.70, 97.84, 98.58, 97.13, 96.59, and 98.28% for AlexNet, VGGNet, P-ResNet, GoogLeNet, MobileNet, DenseNet, ShuffleNet, and EfficientNet, respectively. The highest classification accuracy result was obtained with P-ResNet, and the model loss remained at around 0.01. This model obtained the accuracy of classifications for BaoQiu, ShanCu, XinNuo, LiaoGe, and KouXian varieties, which reached 99.74, 99.68, 99.68, 99.61, and 99.80%, respectively. The experimental results demonstrated that the convolutional neural network model proposed enables the effective classification of maize seeds. It can provide a reference for identifying seeds of other crops and be applied to consumer use and the food industry.

Seed quality affects crop yield and the quality of agricultural products, and traditional identification methods are time-consuming, complex, and irreversibly destructive. This study aims to establish a fast, non-destructive, and effective approach for defect detection in maize seeds based on hyperspectral imaging (HSI) technology combined with deep learning. Raw spectra collected from maize seeds (200 each healthy and worm-eaten) were pre-processed using detrending (DE) and multiple scattering correction (MSC) to highlight the spectral differences between samples. A convolutional neural network architecture (CNN-FES) based on a feature selection mechanism was proposed according to the importance of wavelength in the target classification task. The results show that the subset of 24 feature wavelengths selected by the proposed CNN-FES can capture important feature information in the spectral data more effectively than the conventional successive projections algorithm (SPA) and competitive adaptive reweighted sampling (CARS) algorithms. In addition, a convolutional neural network architecture (CNN-ATM) based on an attentional classification mechanism was designed for one-dimensional spectral data classification and compared with three commonly used machine learning methods, linear discriminant analysis (LDA), random forest (RF), and support vector machine (SVM). The results show that the classification performance of the designed CNN-ATM on the full wavelength does not differ much from the above three methods, and the classification accuracy is above 90% on both the training and test sets. Meanwhile, the accuracy, sensitivity, and specificity of CNN-ATM based on feature wavelength modeling can reach up to 97.50%, 98.28%, and 96.77% at the highest, respectively. The study shows that hyperspectral imaging-based defect detection of maize seed is feasible and effective, and the proposed method has great potential for the processing and analysis of complex hyperspectral data.

In this paper, an adaptive header height control system is designed. Through the influence of the natural frequency, ωn, and damping ratio, ζ, on the system’s dynamic index, the optimal hydraulic cylinder parameters are determined comprehensively. The ground profiling monitoring mechanism and the header height feedback mechanism based on the angle sensor are designed. An integrated electromagnetic proportional valve was installed to replace the original header-controlled, electronically-controlled reversing valve, and a PWM (pulse width modulation) control-simulated counterweight test was performed. The limitation of traditional PID facing the integral saturation state is analysed, and a new EVPIVS-PID algorithm is proposed and simulated. Through the analysis of multiple groups of sample data in the field test, the accuracy of the control system in the header height control and output PWM value is demonstrated. The effectiveness of the EVPIVS-PID control algorithm to change the corresponding PID parameters based on the monitoring operation speed is analysed and demonstrated. Experiments show that the adaptive control system of header height based on ground profiling has a stable control effect. The height error of cutting stubble is not more than 2 cm, which can meet the requirements of a 5–11 km/h harvesting speed in plain areas.

No abstract