Convolutional Neural Network Features Research Articles

Deep machine learning identified fish flesh using multispectral imaging

Food fraud is widespread in the aquatic food market, hence fast and non-destructive methods of identification of fish flesh are needed. In this study, multispectral imaging (MSI) was used to screen flesh slices from 20 edible fish species commonly found in the sea around Yantai, China, by combining identification based on the mitochondrial COI gene. We found that nCDA images transformed from MSI data showed significant differences in flesh splices of the 20 fish species. We then employed eight models to compare their prediction performances based on the hold-out method with 70% training and 30% test sets. Convolutional neural network (CNN), quadratic discriminant analysis (QDA), support vector machine (SVM), and linear discriminant analysis (LDA) models perform well on cross-validation and test data. CNN and QDA achieved more than 99% accuracy on the test set. By extracting the CNN features for optimization, a very high degree of separation was obtained for all species. Furthermore, based on the Gini index in RF, 11 bands were selected as key classification features for CNN, and an accuracy of 98% was achieved. Our study developed a successful pipeline for employing machine learning models (especially CNN) on MSI identification of fish flesh, and provided a convenient and non-destructive method to determine the marketing of fish flesh in the future.

Integrating Convolution and Sparse Coding for Learning Low-Dimensional Discriminative Image Representations.

This work investigates the problem of efficiently learning discriminative low-dimensional (LD) representations of multiclass image objects. We propose a generic end-to-end approach that jointly optimizes sparse dictionary and convolutions for learning LOW-dimensional discriminative image representations, named SparConvLow, taking advantage of convolutional neural networks (CNNs), dictionary learning, and orthogonal projections. The whole learning process can be summarized as follows. First, a CNN module is employed to extract high-dimensional (HD) preliminary convolutional features. Second, to avoid the high computational cost of direct sparse coding on HD CNN features, we learn sparse representation (SR) over a task-driven dictionary in the space with the feature being orthogonally projected. We then exploit the discriminative projection on SR. The whole learning process is consistently treated as an end-to-end joint optimization problem of trace quotient maximization. The cost function is well-defined on the product of the CNN parameters space, the Stiefel manifold, the Oblique manifold, and the Grassmann manifold. By using the explicit gradient delivery, the cost function is optimized via a geometrical stochastic gradient descent (SGD) algorithm along with the chain rule and the backpropagation. The experimental results show that the proposed method can achieve a highly competitive performance with the state-of-the-art (SOTA) image classification, object categorization, and face recognition methods, under both supervised and semi-supervised settings. The code is available at https://github.com/MVPR-Group/SparConvLow.

MSCT-UNET: multi-scale contrastive transformer within U-shaped network for medical image segmentation

Objective. Automatic mutli-organ segmentation from anotomical images is essential in disease diagnosis and treatment planning. The U-shaped neural network with encoder–decoder has achieved great success in various segmentation tasks. However, a pure convolutional neural network (CNN) is not suitable for modeling long-range relations due to limited receptive fields, and a pure transformer is not good at capturing pixel-level features. Approach. We propose a new hybrid network named MSCT-UNET which fuses CNN features with transformer features at multi-scale and introduces multi-task contrastive learning to improve the segmentation performance. Specifically, the multi-scale low-level features extracted from CNN are further encoded through several transformers to build hierarchical global contexts. Then the cross fusion block fuses the low-level and high-level features in different directions. The deep-fused features are flowed back to the CNN and transformer branch for the next scale fusion. We introduce multi-task contrastive learning including a self-supervised global contrast learning and a supervised local contrast learning into MSCT-UNET. We also make the decoder stronger by using a transformer to better restore the segmentation map. Results. Evaluation results on ACDC, Synapase and BraTS datasets demonstrate the improved performance over other methods compared. Ablation study results prove the effectiveness of our major innovations. Significance. The hybrid encoder of MSCT-UNET can capture multi-scale long-range dependencies and fine-grained detail features at the same time. The cross fusion block can fuse these features deeply. The multi-task contrastive learning of MSCT-UNET can strengthen the representation ability of the encoder and jointly optimize the networks. The source code is publicly available at: https://github.com/msctunet/MSCT_UNET.git.

HCA-former: Hybrid Convolution Attention Transformer for 3D Medical Image Segmentation

In recent years, Transformers have achieved success in the field of medical image segmentation due to their outstanding capability to model long-range dependencies. However, many existing segmentation methods only use Transformer as an auxiliary module to capture global context information in images, limiting the potential of the Transformers. Additionally, self-attention mechanism within the Transformers can lead to attention collapse issues, thus triggering semantic gap between the encoder and decoder. Furthermore, most networks have difficulties in effectively handling multi-scale and multi-channel feature information. To address the above problems, we propose a hybrid Convolutional Neural Networks (CNNs) and Transformers method for medical image segmentation (HCA-Former). We design a local multi-channel attention block (LMCA) to effectively combine the features of CNN and Transformers, enabling multi-channel information extraction and interaction. Using the Double-Former Block (DFB) alleviates the semantic gap between the encoder and decoder, restoring more detailed information. Moreover, the utilization of the global multi-scale attention block (GMSA) can establish information interaction among multi-scale targets, thereby enhancing generalization capability of the model. To validate the effectiveness of our approach, we evaluate the proposed method on three challenging tasks: the MICCAI 2015 Multi-Image Abdominal Marker Challenge (Synapse), Automated Cardiac Diagnosis Dataset (ACDC), and Medical Segmentation Decathlon Brain Tumor Segmentation (MSD brain tumor). Extensive experiments demonstrate that our HCA-Former achieved competitive or better performance than state-of-the-art approaches for 3D medical image segmentation.

Hybrid rat swarm hunter prey optimization trained deep learning for network intrusion detection using CNN features

In recent years, the emerging growth of various network technologies, namely the cloud computing, 5 G networks, Internet of Things (IoT), and so on resulted in the createion of huge quantity of data and creates challenges to the security of networks. The classical firewalls failed to meet various requirements of the current security of the networks due to the increasing number of network attacks. Hence, an optimization-enabled deep learning model Rat Swarm Hunter Prey Optimization-Deep Maxout Network (RSHPO-DMN) technique is designed to effectively handle various network threats. The Z-score data normalization is applied for data pre-processing initially and by considering chord distance, the data is transformed into usable formats. The transformed data are extracted using the Convolutional Neural Network (CNN) feature and the extracted feature is converted into vector format for network intrusion detection process. The DMN is used for intrusion detection in networks and the designed RSHPO model is used to boost the intrusion detection rate exhibited in the classifier. The RSHPO-DMN model achieved superior performance under different evaluation indicators with accuracy of 90.88 %, precision of 93.58 %, recall of 96.54 %, and f1-score of 95.04 % respectively than other prevailing intrusion detection approaches.

Research on high proportion of clean energy grid-connected oscillation risk prediction technology based on CNN and trend feature analysis

Oscillations, commonly known as a universal, propagative, and intricate event in the new power system, often give rise to generator tripping and load shedding, not only adversely affecting the power flow limit and the power angle stability but also posing threats to the lines of defense for stability and protection. Traditionally, emphasis has been laid on post-fault oscillation management, an emergency measure to deal with the impact and damage that have already affected the power grid. As such, this paper focuses on an oscillation prediction technique to detect oscillation energy early and intervene proactively to prevent further faults. This technique effectively lessens the damage caused by impacts and disconnects to the power grid. Firstly, this paper proposes the concept of disturbance power density and establishes the correlation between disturbance energy and the time domain, thereby exploring a method for evaluating the pattern of electrical quantities before power system oscillation. Secondly, it speeds up the time it takes to detect faults by catching nuances of voltage-current phase angle and impedance. Lastly, it puts forward a technique to cope with the intricacy and variety of power grid equipment using the convolutional neural network (CNN). This technique incorporates an integrated attention mechanism within a one-dimensional CNN model to capture the implicit mapping between voltage, active power, and reactive power at any time in the power system. This enables the model to self-learn multi-device characteristics and enhances the possibility of using theory in practical ways. Moreover, practical case studies also show that the prediction technique proposed in this paper can effectively issue warnings eight minutes before the occurrence of oscillation.

Zero-Shot Traffic Sign Recognition Based on Midlevel Feature Matching

Traffic sign recognition is a complex and challenging yet popular problem that can assist drivers on the road and reduce traffic accidents. Most existing methods for traffic sign recognition use convolutional neural networks (CNNs) and can achieve high recognition accuracy. However, these methods first require a large number of carefully crafted traffic sign datasets for the training process. Moreover, since traffic signs differ in each country and there is a variety of traffic signs, these methods need to be fine-tuned when recognizing new traffic sign categories. To address these issues, we propose a traffic sign matching method for zero-shot recognition. Our proposed method can perform traffic sign recognition without training data by directly matching the similarity of target and template traffic sign images. Our method uses the midlevel features of CNNs to obtain robust feature representations of traffic signs without additional training or fine-tuning. We discovered that midlevel features improve the accuracy of zero-shot traffic sign recognition. The proposed method achieves promising recognition results on the German Traffic Sign Recognition Benchmark open dataset and a real-world dataset taken from Sapporo City, Japan.

Feature selection based on unsupervised clustering evaluation for predicting neoadjuvant chemoradiation response for patients with locally advanced rectal cancer

Accurate response prediction allows for personalized cancer treatment of locally advanced rectal cancer (LARC) with neoadjuvant chemoradiation. In this work, we designed a convolutional neural network (CNN) feature extractor with switchable 3D and 2D convolutional kernels to extract deep learning features for response prediction. Compared with radiomics features, convolutional kernels may adaptively extract local or global image features from multi-modal MR sequences without the need of feature predefinition. We then developed an unsupervised clustering based evaluation method to improve the feature selection operation in the feature space formed by the combination of CNN features and radiomics features. While normal process of feature selection generally includes the operations of classifier training and classification execution, the process needs to be repeated many times after new feature combinations were found to evaluate the model performance, which incurs a significant time cost. To address this issue, we proposed a cost effective process to use a constructed unsupervised clustering analysis indicator to replace the classifier training process by indirectly evaluating the quality of new found feature combinations in feature selection process. We evaluated the proposed method using 43 LARC patients underwent neoadjuvant chemoradiation. Our prediction model achieved accuracy, area-under-curve (AUC), sensitivity and specificity of 0.852, 0.871, 0.868, and 0.735 respectively. Compared with traditional radiomics methods, the prediction models (AUC = 0.846) based on deep learning-based feature sets are significantly better than traditional radiomics methods (AUC = 0.714). The experiments also showed following findings: (1) the features with higher predictive power are mainly from high-order abstract features extracted by CNN on ADC images and T2 images; (2) both ADC_Radiomics and ADC_CNN features are more advantageous for predicting treatment responses than the radiomics and CNN features extracted from T2 images; (3) 3D CNN features are more effective than 2D CNN features in the treatment response prediction. The proposed unsupervised clustering indicator is feasible with low computational cost, which facilitates the discovery of valuable solutions by highlighting the correlation and complementarity between different types of features.

A Multimodal Multilevel Converged Attention Network for Hand Gesture Recognition With Hybrid sEMG and A-Mode Ultrasound Sensing.

Gesture recognition based on surface electromyography (sEMG) has been widely used in the field of human-machine interaction (HMI). However, sEMG has limitations, such as low signal-to-noise ratio and insensitivity to fine finger movements, so we consider adding A-mode ultrasound (AUS) to enhance the recognition impact. To explore the influence of multisource sensing data on gesture recognition and better integrate the features of different modules. We proposed a multimodal multilevel converged attention network (MMCANet) model for multisource signals composed of sEMG and AUS. The proposed model extracts the hidden features of the AUS signal with a convolutional neural network (CNN). Meanwhile, a CNN-LSTM (long-short memory network) hybrid structure extracts some spatial-temporal features from the sEMG signal. Then, two types of CNN features from AUS and sEMG are spliced and transmitted to a transformer encoder to fuse the information and interact with sEMG features to produce hybrid features. Finally, the classification results are output employing fully connected layers. Attention mechanisms are used to adjust the weights of feature channels. We compared MMCANet's feature extraction and classification performance with that of manually extracted sEMG-AUS features using four traditional machine-learning (ML) algorithms. The recognition accuracy increased by at least 5.15%. In addition, we tried deep learning (DL) methods with CNN on single modals. The experimental results showed that the proposed model improved 14.31% and 3.80% over the CNN method with single sEMG and AUS, respectively. Compared with some state-of-the-art fusion techniques, our method also achieved better results.

Fault diagnosis for PV arrays considering dust impact based on transformed graphical features of characteristic curves and convolutional neural network with CBAM modules

Various faults can occur during the operation of photovoltaic (PV) arrays, and both dust-affected operating conditions and various diode configurations complicate the faults. However, current methods for fault diagnosis based on I-V characteristic curves usually do not effectively use all the distinguishable information contained in the I-V curves or often rely on calibrating the field characteristic curves to standard test conditions (STC). It is difficult to apply these methods in practice and accurately identify multiple complex faults with similarities in different blocking diode configurations of PV arrays under the influence of dust. Therefore, a novel fault-diagnosis method for PV arrays that considers the impact of dust is proposed. In the pre-processing stage, the Isc-Voc normalized Gramian angular difference field (GADF) method is presented, which normalizes and transforms the resampled PV array characteristic curves from the field, including I-V and P-V, to obtain transformed graphical feature matrices. Subsequently, in the fault diagnosis stage, the convolutional neural network (CNN) model with convolutional block attention modules (CBAM) is designed to classify faults, which identifies complex fault types from the transformed graphical matrices containing complete discriminative fault information. In addition, the performances of different graphical feature transformation and CNN-based classification methods are compared using case studies. The results indicate that the developed method for PV arrays with different blocking diode configurations under various operating conditions has high fault diagnosis accuracy and reliability.

Combined model integrating deep learning, radiomics, and clinical data to classify lung nodules at chest CT.

The study aimed to develop a combined model that integrates deep learning (DL), radiomics, and clinical data to classify lung nodules into benign or malignant categories, and to further classify lung nodules into different pathological subtypes and Lung Imaging Reporting and Data System (Lung-RADS) scores. The proposed model was trained, validated, and tested using three datasets: one public dataset, the Lung Nodule Analysis 2016 (LUNA16) Grand challenge dataset (n = 1004), and two private datasets, the Lung Nodule Received Operation (LNOP) dataset (n = 1027) and the Lung Nodule in Health Examination (LNHE) dataset (n = 1525). The proposed model used a stacked ensemble model by employing a machine learning (ML) approach with an AutoGluon-Tabular classifier. The input variables were modified 3D convolutional neural network (CNN) features, radiomics features, and clinical features. Three classification tasks were performed: Task 1: Classification of lung nodules into benign or malignant in the LUNA16 dataset; Task 2: Classification of lung nodules into different pathological subtypes; and Task 3: Classification of Lung-RADS score. Classification performance was determined based on accuracy, recall, precision, and F1-score. Ten-fold cross-validation was applied to each task. The proposed model achieved high accuracy in classifying lung nodules into benign or malignant categories in LUNA 16 with an accuracy of 92.8%, as well as in classifying lung nodules into different pathological subtypes with an F1-score of 75.5% and Lung-RADS scores with an F1-score of 80.4%. Our proposed model provides an accurate classification of lung nodules based on the benign/malignant, different pathological subtypes, and Lung-RADS system.

Fusion of Attention-Based Convolution Neural Network and HOG Features for Static Sign Language Recognition

The deaf and hearing-impaired community expresses their emotions, communicates with society, and enhances the interaction between humans and computers using sign language gestures. This work presents a strategy for efficient feature extraction that uses a combination of two different methods that are the convolutional block attention module (CBAM)-based convolutional neural network (CNN) and standard handcrafted histogram of oriented gradients (HOG) feature descriptor. The proposed framework aims to enhance accuracy by extracting meaningful features and resolving issues like rotation, similar hand orientation, etc. The HOG feature extraction technique provides a compact feature representation that signifies meaningful information about sign gestures. The CBAM attention module is incorporated into the structure of CNN to enhance feature learning using spatial and channel attention mechanisms. Then, the final feature vector is formed by concatenating these features. This feature vector is provided to the classification layers to predict static sign gestures. The proposed approach is validated on two publicly available static Massey American Sign Language (ASL) and Indian Sign Language (ISL) databases. The model’s performance is evaluated using precision, recall, F1-score, and accuracy. Our proposed methodology achieved 99.22% and 99.79% accuracy for the ASL and ISL datasets. The acquired results signify the efficiency of the feature fusion and attention mechanism. Our network performed better in accuracy compared to the earlier studies.

Mechanomyography signals pattern recognition in hand movements using swarm intelligence algorithm optimized support vector machine based on acceleration sensors

On the basis of extracting mechanomyography (MMG) signal features, the classification of hand movements has certain application values in human-machine interaction systems and wearable devices. In this paper, pattern recognition of hand movements based on MMG signal is studied with swarm intelligence algorithms introduced to optimize support vector machine (SVM). Time domain (TD) features, wavelet packet node energy (WPNE) features, frequency domain (FD) features, convolution neural network (CNN) features were extracted from each channel to constitute different feature sets. Three novel swarm intelligence algorithms (i.e., bald eagle search (BES), sparrow search algorithm (SSA), grey wolf optimization (GWO)) optimized SVM is proposed to train the models and recognition of hand movements are tested for each MMG feature extraction method. Using GWO as the optimization algorithm, time consumption is less than using the other two swarm algorithms. Using GWO with TD+FD features can obtain the classification accuracy of 93.55 %, which is higher than other methods while using CNN to extract features can be independent of domain knowledge. The results confirm GWO-SVM with TD + FD features is superior to some other methods in the classification problem for tiny samples based on MMG.

CCF-GNN: A Unified Model Aggregating Appearance, Microenvironment, and Topology for Pathology Image Classification.

Pathology images contain rich information of cell appearance, microenvironment, and topology features for cancer analysis and diagnosis. Among such features, topology becomes increasingly important in analysis for cancer immunotherapy. By analyzing geometric and hierarchically structured cell distribution topology, oncologists can identify densely-packed and cancer-relevant cell communities (CCs) for making decisions. Compared to commonly-used pixel-level Convolution Neural Network (CNN) features and cell-instance-level Graph Neural Network (GNN) features, CC topology features are at a higher level of granularity and geometry. However, topological features have not been well exploited by recent deep learning (DL) methods for pathology image classification due to lack of effective topological descriptors for cell distribution and gathering patterns. In this paper, inspired by clinical practice, we analyze and classify pathology images by comprehensively learning cell appearance, microenvironment, and topology in a fine-to-coarse manner. To describe and exploit topology, we design Cell Community Forest (CCF), a novel graph that represents the hierarchical formulation process of big-sparse CCs from small-dense CCs. Using CCF as a new geometric topological descriptor of tumor cells in pathology images, we propose CCF-GNN, a GNN model that successively aggregates heterogeneous features (e.g., appearance, microenvironment) from cell-instance-level, cell-community-level, into image-level for pathology image classification. Extensive cross-validation experiments show that our method significantly outperforms alternative methods on H&E-stained and immunofluorescence images for disease grading tasks with multiple cancer types. Our proposed CCF-GNN establishes a new topological data analysis (TDA) based method, which facilitates integrating multi-level heterogeneous features of point clouds (e.g., for cells) into a unified DL framework.

An End-to-End Deep Hybrid Autoencoder Based Method for Single-Cell RNA-Seq Data Analysis.

Single-cell RNA sequencing technology provides powerful support for researchers to understand the complex mechanisms of cells at the single-cell level. Due to the high sparsity, technical noise, and computational complexity of single-cell transcriptome data, the existing data analysis methods are unable to effectively extract the fine-grained characteristics of scRNA-seq data, resulting in inaccurately analyze the heterogeneity of the individual cell from a great quantity of cell mixtures. To address these shortcomings, we proposed an end-to-end analysis method called dhaSCA, which integrates the Graph convolutional neural network (GCN) feature learning and downstream tasks such as classification and imputation into a unified deep learning manner. dhaSCA uses hybrid GCN-MLP deep autoencoder and to capture structural information between cells, and learn the low dimensional cell representation. It also introduces downstream tasks as constraints to guide the model to learn more accurate cell features. We conducted various experiments to evaluate the performance of dhaSCA based on eight real RNA-Seq datasets, including classification, imputation, clustering, and visualization. The results show that dhaSCA outperforms other state-of-the-art methods in these downstream tasks. Therefore, dhaSCA is able to obtain a richer representation of cells, and provides strong support for efficient analysis of single-cell data.

Covid CNN–SVM: Automated Categorization of Novel Coronavirus Disease from X-Ray Images

In the computer-aided diagnostic (CAD) system, automated Coronavirus infection disclosure plays a crucial role in early identifying positive patients to prevent the disease from spreading further. The advent of algorithms for deep learning and machine learning has tackled classification tasks with promising results, especially in classifying images. However, the small size of the databases for medical images is a limitation associated with train deep neural networks. We use a combination of convolutional neural network (CNN) features and a support vector machine (SVM) for X-ray image classification to overcome this problem. This research work used CNN methods to extract features from 1,338 Chest X-ray frontal view image data. An SVM is used with CNN features to classify images in two classes: COVID-19 and Normal cases for enhanced performance. We conducted and evaluated our experiments on several public databases, which have been used in the recently published articles. The performance of the proposed method revealed accuracy, AUC, sensitivity, specificity of 0.995, 0.999, and 0.995 for classification, respectively. The high performance of the detection system achieved in this research reveals the effectiveness of deep features and the machine learning classifier approach for detecting COVID-19 cases using X-ray images. This would be extremely helpful in accelerating disease diagnosis with the available resources.

Brain tumor image segmentation based on improved FPN

PurposeAutomatic segmentation of brain tumors by deep learning algorithm is one of the research hotspots in the field of medical image segmentation. An improved FPN network for brain tumor segmentation is proposed to improve the segmentation effect of brain tumor.Materials and methodsAiming at the problem that the traditional full convolutional neural network (FCN) has weak processing ability, which leads to the loss of details in tumor segmentation, this paper proposes a brain tumor image segmentation method based on the improved feature pyramid networks (FPN) convolutional neural network. In order to improve the segmentation effect of brain tumors, we improved the model, introduced the FPN structure into the U-Net structure, captured the context multi-scale information by using the different scale information in the U-Net model and the multi receptive field high-level features in the FPN convolutional neural network, and improved the adaptability of the model to different scale features.ResultsPerformance evaluation indicators show that the proposed improved FPN model has 99.1% accuracy, 92% DICE rating and 86% Jaccard index. The performance of the proposed method outperforms other segmentation models in each metric. In addition, the schematic diagram of the segmentation results shows that the segmentation results of our algorithm are closer to the ground truth, showing more brain tumour details, while the segmentation results of other algorithms are smoother.ConclusionsThe experimental results show that this method can effectively segment brain tumor regions and has certain generalization, and the segmentation effect is better than other networks. It has positive significance for clinical diagnosis of brain tumors.

Gannet devil optimization-based deep learning for skin lesion segmentation and identification

Skin lesion identification is a way of identifying abnormal skin tissues from normal skin. The affected area of the skin are segmented in the skin lesion segmentation process, which helps to improve the identification task. In this paper, the skin lesion segmentation and identification process are completed through efficient deep-learning networks. Here, the Wiener filter is employed for preprocessing and is used to cleans up the noise present in the image. The UNet++ model completes the skin lesion segmentation and it separates the affected region from the noise-free images. By considering the segmented portion, the prevalent features namely Convolutional Neural Network (CNN) features, Local Optimal Oriented Pattern (LOOP), statistical features, Gray-Level Co-Occurrence Matrix (GLCM) features, Local ternary pattern (LTP), and Shape Local Binary Texture (SLBT) features are mined out. Moreover, the LeNet model completes the skin lesion segmentation and the weight of both UNet++ and LeNet is tuned by the devised Gannet Devil Optimization Algorithm (GDO), which is the incorporation of Gannet Optimization Algorithm (GOA) and Tasmanian Devil Optimization algorithm (TDO). To segment and classify the lesions from the medical images, the dataset SIIM-ISIC Melanoma Identification data is taken into consideration. Besides, the experimental result indicates the GDO-LeNet provided improved performance regarding the accuracy, True Positive Rate (TPR), True Negative Rate (TNR), positive predictive value (PPV), and Negative predictive value (NPV) of 0.924, 0.925, 0.937, 0.896 and 0.890.

Research on the Strawberry Recognition Algorithm Based on Deep Learning

In view of the time-consuming and laborious manual picking and sorting of strawberries, the direct impact of image recognition accuracy on automatic picking and the rapid development of deep learning(DL), a Faster Regions with Convolutional Neural Network features (R-CNN) strawberry recognition method that combines Mixup data augmentation, a ResNet(Residual Network)50 backbone feature extraction network and a Soft-NMS (Non-Maximum Suppression) algorithm, named the MRS Faster R-CNN, is proposed. In this paper, the transfer learning backbone feature extraction network VGG (Visual Geometry Group) 16 and ResNet50 are compared, and the superior ResNet50 is selected as the backbone network of MRS Faster R-CNN. The data augmentation method of Mixup image fusion is used to improve the learning and generalization ability of the model. The redundant bboxes (bounding boxes) are removed through Soft-NMS to obtain the best region proposal. The freezing phase is added to the training process, effectively reducing the occupation of video memory and shortening the training time. After experimental verification, the optimized model improved the AP (Average Precision) values of mature and immature strawberries by 0.26% and 5.34%, respectively, and the P(Precision) values by 0.81% and 6.34%, respectively, compared to the original model (R Faster R-CNN). Therefore, the MRS Faster R-CNN model proposed in this paper has great potential in the field of strawberry recognition and maturity classification and improves the recognition rate of small fruit and overlapping occluded fruit, thus providing an excellent solution for mechanized picking and sorting.

A SENTIMENT ANALYSIS MODEL FOR ELECTROENCEPHALOGRAM SIGNALS OF STUDENTS IN UNIVERSITIES USING A CONVOLUTIONAL NEURAL NETWORK AND SUPPORT VECTOR MACHINE MODELS

Sentiment analysis in teaching evaluation has significant implications. By analyzing students’ sentiments toward instructors, educational institutions can gain valuable insights into teaching effectiveness. These data can guide curriculum development, instructional improvements, and faculty training initiatives. Positive sentiment indicates effective teaching methods, engagement, and student satisfaction; negative sentiment flags areas that need attention. Sentiment analysis can help identify patterns, trends, and outliers, aiding in targeted interventions and personalized support. It also enables comparisons across different courses, instructors, and departments. However, it is crucial to ensure the accuracy and fairness of sentiment analysis algorithms, considering potential biases and the contextual nature of the feedback. This study proposes a sentiment classification model CNN–SVM that combines a convolutional neural network (CNN) and a support vector machine (SVM). Taking students majoring in art in comprehensive colleges and universities as the research object, by collecting the electroencephalogram (EEG) signals of students during teaching evaluation. CNN–SVM is used as the emotional analysis model to obtain the emotional analysis of teaching evaluation results. EEG is a typical physiological signal, and data based on this signal can more truly reflect student emotions. The adaptive CNN feature extraction function and the super generalization classification performance of SVM can reduce the individual differences and data noise between data, thereby improving sentiment classification performance. The experimental results demonstrate that using technology to analyze sentiment can assist educational institutions in more properly comprehending the feedback and opinions of students on instruction. With regard to sentiment analysis, the CNN–SVM method that is derived to produce the fusion algorithm has solid performance.