CNN Classifier Research Articles

SAM-ResNet50: A Deep Learning Model for the Identification and Classification of Drought Stress in the Seedling Stage of Betula luminifera

Betula luminifera, an indigenous hardwood tree in South China, possesses significant economic and ecological value. In view of the current severe drought situation, it is urgent to enhance this tree’s drought tolerance. However, traditional artificial methods fall short of meeting the demands of breeding efforts due to their inefficiency. To monitor drought situations in a high-throughput and automatic approach, a deep learning model based on phenotype characteristics was proposed to identify and classify drought stress in B. luminifera seedlings. Firstly, visible-light images were obtained from a drought stress experiment conducted on B. luminifera shoots. Considering the images’ characteristics, we proposed an SAM-CNN architecture by incorporating spatial attention modules into classical CNN models. Among the four classical CNNs compared, ResNet50 exhibited superior performance and was, thus, selected for the construction of the SAM-CNN. Subsequently, we analyzed the classification performance of the SAM-ResNet50 model in terms of transfer learning, training from scratch, model robustness, and visualization. The results revealed that SAM-ResNet50 achieved an accuracy of 1.48% higher than that of ResNet50, at 99.6%. Furthermore, there was a remarkable improvement of 18.98% in accuracy, reaching 82.31% for the spatial transform images generated from the test set images by applying movement and rotation for robustness testing. In conclusion, the SAM-ResNet50 model achieved outstanding performance, with 99.6% accuracy and realized high-throughput automatic monitoring based on phenotype, providing a new perspective for drought stress classification and technical support for B. luminifera-related breeding work.

Detecting and Localizing Wireless Spoofing Attacks on the Internet of Medical Things

This paper proposes a hybrid approach using design science research to identify rogue RF transmitters and locate their targets. We engineered a framework to identify masquerading attacks indicating the presence of multiple adversaries posing as a single node. We propose a methodology based on spatial correlation calculated from received signal strength (RSS). To detect and mitigate wireless spoofing attacks in IoMT environments effectively, the hybrid approach combines spatial correlation analysis, Deep CNN classification, Elliptic Curve Cryptography (ECC) encryption, and DSRM-powered attack detection enhanced (DADE) detection and localization (DAL) frameworks. A deep neural network (Deep CNN) was used to classify trusted transmitters based on Python Spyder3 V5 and ECC encrypted Hack RF Quadrature Signals (IQ). For localizing targets, this paper also presents DADE and DAL frameworks implemented on Eclipse Java platforms. The hybrid approach relies on spatial correlation based on signal strength. Using the training methods of Deep CNN1, Deep CNN2, and Long Short-Term Memory (LSTM), it was possible to achieve accuracies of 98.88%, 95.05%, and 96.60% respectively.

Center-Highlighted Multiscale CNN for Classification of Hyperspectral Images

Hyperspectral images (HSIs) capture a wide range of spectral features across multiple bands of light, from visible to near-infrared. Hyperspectral image classification technology enables researchers to accurately identify and analyze the composition and distribution of surface materials. Current mainstream deep learning methods typically use block sampling to capture spatial features for the model. However, this approach can affect classification results due to the influence of neighboring features within the sample block. To improve the model’s focus on the center of the sampling block, this study proposes a center highlight with multiscale CNN for hyperspectral image classification (CHMSC). The network utilizes an automatic channel selector (Auto-CHS) to fully consider every channel feature and capture the correlation between the channels. Then, CHMSC enhances the model’s ability to concentrate on the central features of the sampling block utilizing structures such as the center highlight. Finally, before outputting the prediction results, an SENet is employed to further refine the features and learn associate interactions between different scales of spatial features and spectral features. Experimental results from three hyperspectral datasets validate the effectiveness of the proposed method. Specifically, when 15 samples from each class are selected for training, CHMSC achieves the highest overall accuracy (OA) of 90.05%, 92.78%, and 90.15% on the three datasets, outperforming other methods with increases of more than 3.11%, 1.8%, and 2.01% in OA, respectively.

PBNet: Combining Transformer and CNN in Passport Background Texture Printing Image Classification

Passport background texture classification has always been an important task in border checks. Current manual methods struggle to achieve satisfactory results in terms of consistency and stability for weakly textured background images. For this reason, this study designs and develops a CNN and Transformer complementary network (PBNet) for passport background texture image classification. We first design two encoders by Transformer and CNN to produce complementary features in the Transformer and CNN domains, respectively. Then, we cross-wisely concatenate these complementary features to propose a feature enhancement module (FEM) for effectively blending them. In addition, we introduce focal loss to relieve the overfitting problem caused by data imbalance. Experimental results show that our PBNet significantly surpasses the state-of-the-art image segmentation models based on CNNs, Transformers, and even Transformer and CNN combined models designed for passport background texture image classification.

Measuring Student Engagement through Behavioral and Emotional Features Using Deep-Learning Models

Students’ behavioral and emotional engagement in the classroom environment may reflect the students’ learning experience and subsequent educational outcomes. The existing research has overlooked the measurement of behavioral and emotional engagement in an offline classroom environment with more students, and it has not measured the student engagement level in an objective sense. This work aims to address the limitations of the existing research and presents an effective approach to measure students’ behavioral and emotional engagement and the student engagement level in an offline classroom environment during a lecture. More precisely, video data of 100 students during lectures in different offline classes were recorded and pre-processed to extract frames with individual students. For classification, convolutional-neural-network- and transfer-learning-based models including ResNet50, VGG16, and Inception V3 were trained, validated, and tested. First, behavioral engagement was computed using salient features, for which the self-trained CNN classifier outperformed with a 97%, 91%, and 83% training, validation, and testing accuracy, respectively. Subsequently, the emotional engagement of the behaviorally engaged students was computed, for which the ResNet50 model surpassed the others with a 95%, 90%, and 82% training, validation, and testing accuracy, respectively. Finally, a novel student engagement level metric is proposed that incorporates behavioral and emotional engagement. The proposed approach may provide support for improving students’ learning in an offline classroom environment and devising effective pedagogical policies.

Uncertainty-Aware Seismic Signal Discrimination using Bayesian Convolutional Neural Networks

Seismic signal classification plays a crucial role in mitigating the impact of seismic events on human lives and infrastructure. Traditional methods in seismic hazard assessment often overlook the inherent uncertainties associated with the prediction of this complex geological phenomenon. This work introduces a probabilistic framework that leverages Bayesian principles to model and quantify uncertainty in seismic signal classification by applying a Bayesian Convolutional Neural Network (BCNN). The BCNN was trained on a dataset that comprises waveforms detected in the Southern California region and achieved an accuracy of 99.1%. Monte Carlo Sampling subsequently creates a 95% prediction interval for probabilities that considers epistemic and aleatoric uncertainties. The ability to visualize both aleatoric and epistemic uncertainties provides decision-makers with information to determine the reliability of seismic signal classifications. Further, the use of Bayesian CNN for seismic signal classification provides a more robust foundation for decision-making and risk assessment in earthquake-prone regions.

Machine learning based Parkinson’s disease detection and progression evaluation using gray and white matter segmentation from 3D MRI

Parkinson’s disease (PD) is a movement-related neurological condition caused by the death of brain nerve cells that produce dopamine. T1 MR images were obtained from the Parkinson’s Progression Markers Initiative (PPMI) database. Data was collected at baseline, and at 48 months duration. SPM12 software was used to segment gray matter (GM) and white matter (WM) from the MR images. For the classification of PD, CNN and machine learning (ML) classifiers were used to train on the segmented GM and WM. The findings demonstrated that utilizing the segmented GM and WM obtained from MR images at 48 months had a better predictive ability than the data obtainment at the baseline. CNNs did not perform as well as the conventional ML algorithms, especially for the baseline data. This result is most likely due to the smaller dataset available for training the CNNs, as CNNs normally require more data for best performance. Overall, machine learning algorithms were able to distinguish between people with Parkinson’s disease and healthy controls by analyzing GM and WM in brain scans. The classifiers trained at different stages demonstrated varying degrees of accuracy, with the predictive accuracy for the 48-month dataset surpassing that of the baseline data by a significant margin. The accuracy rate for GM was 65.78% at baseline and improved significantly to 92.59% at 48 months. Similarly, the accuracy rate for WM was 60.52% at baseline and improved to 88.89% at 48 months.

Enhancing alzheimer’s diagnosis through optimized brain lesion classification in MRI with attention-driven grid feature fusion

This paper explores cognitive interface technology, aiming to tackle current challenges and shed light on the prospects of brain-computer interfaces (BCIs). It provides a comprehensive examination of their transformative impact on medical technology and patient well-being. Specifically, this study contributes to addressing challenges in classifying brain lesion images arising from the complex nature of lesions and limitations of traditional deep learning approaches. It introduces advanced feature fusion models that leverage deep learning algorithms, including the African vulture optimization (AVO) algorithm. These models integrate informative features from multiple pre-trained networks and employ innovative fusion techniques, including the attention-driven grid feature fusion (ADGFF) model. The ADGFF model incorporates an attention mechanism based on the optimized weights obtained using AVO. The objective is to improve the overall accuracy by providing fine-grained control over different regions of interest in the input image through a grid-based technique. This grid-based technique divides the image into vertical and horizontal grids, simplifying the exemplar feature generation process without compromising performance. Experimental results demonstrate that the proposed feature fusion strategies consistently outperform individual pre-trained models in terms of accuracy, sensitivity, specificity, and F1-score. The optimized feature fusion strategies, particularly the GRU-ADGFF model, further enhance classification performance, outperforming CNN and RNN classifiers. The learning progress analysis shows convergence, indicating the effectiveness of the feature fusion strategies in capturing lesion patterns. AUC-ROC curves highlight the superior discriminatory capabilities of the ADGFF-AVO strategy. Five-fold cross-validation is employed to assess the performance of the proposed models, demonstrating their accuracy, and few other accuracy-based measures. The GRU-ADGFF model optimized with AVO consistently achieves high accuracy, sensitivity, and AUC values, demonstrating its effectiveness and generalization capability. The GRU-ADGFF model also outperforms the majority voting ensemble technique in terms of accuracy and discriminative ability. Additionally, execution time analysis reveals good scalability and resource utilization of the proposed models. The Friedman rank test confirms significant differences in classifier performance, with the GRU-ADGFF model emerging as the top-performing method across different feature fusion strategies and optimization algorithms.

Screening diabetic retinopathy and exudates in retinal images using dual functional convolutional neural networks

Exudates are the abnormal pattern lesions in Diabetic Retinopathy (DR) images, which is the primary cause of DR in diabetic patients. Therefore, its detection process is essential for further severity estimation of DR images. Automated detection systems’ robustness can be hampered by training datasets that are not representative. In this paper, the DR image is detected and classified using the proposed Dual Functional Convolutional Neural Networks (DFCNN). This proposed system consists of a noise detection and removal subblock, enhancement subblock, Gabor transform subblock, Feature computation subblock, and the proposed CNN classification architecture subblock. This proposed methodology is divided into two modules: a training module and a testing module. In the training module, the proposed DFCNN classification architecture is used to train both the abnormal case (DR case) and the healthy case (HCA) retinal images. In the testing module, the test source retinal image is classified as either a beneficial or DR case retinal image. From the classified DR case retinal image, the exudates are identified through the segmentation process in this research work. This developed DR and exudates detection system stated in this paper are evaluated on DRIVE, and DIARETDB1 retinal images and the experimental results are conducted with respect to various performance metrics in this paper. Once the suggested methodology was compared to other classification approaches, the proposed method’s accuracy was found to be greater (99.4 %).

Hybrid DL Models for Improved Accuracy in Diagnosing Chronic Obstructive Pulmonary Disease

Chronic Obstructive Pulmonary Disease (COPD) is a common respiratory disorder marked by enduring airflow obstruction, leading to considerable illness and death rates. Timely and precise diagnosis is essential for proper management and treatment. In this study, we present a novel hybrid deep learning (DL) model leveraging an Autoencoder-GAN (Generative Adversarial Network) architecture to improve the accuracy of COPD diagnosis. Our approach incorporates a cutting-edge preprocessing method, Adaptive Histogram Equalization with Contrast Limited Adaptive Histogram Equalization (CLAHE), to enhance the contrast and detail in CXR images, facilitating more precise feature extraction. The proposed Autoencoder-GAN Hybrid Model significantly outperforms traditional models, achieving an impressive accuracy of 98.3%. By enhancing image quality and focusing on key features through CLAHE preprocessing, our model is able to better distinguish between healthy and COPD-affected lungs. We compared the performance of our model with standard DL models, including CNN, SVM and Random Forest Classifiers, demonstrating superior results across various evaluation metrics. This study highlights the potential of advanced DL techniques and innovative preprocessing methods to enhance the accuracy of COPD diagnosis, offering a promising tool for healthcare professionals in the early detection and management of this chronic disease.

Generation of Rule-Based Explanations of CNN Classifiers Using Regional Features

Although Deep Learning networks generally outperform traditional machine learning approaches based on tailored features, they often lack explainability. To address this issue, numerous methods have been proposed, particularly for image-related tasks such as image classification or object segmentation. These methods generate a heatmap that visually explains the classification problem by identifying the most important regions for the classifier. However, these explanations remain purely visual. To overcome this limitation, we introduce a novel CNN explainability method that identifies the most relevant regions in an image and generates a decision tree based on meaningful regional features, providing a rule-based explanation of the classification model. We evaluated the proposed method on a synthetic blob’s dataset and subsequently applied it to two cell image classification datasets with healthy and pathological patterns.

A Robust Solution for Tea Leaf Classification using the SMOTE and Soft Voting Classifier

Background/Objectives: The purpose of the proposed work is to identify how a tea leaf is conceptualized. This further enables the categorization of leaves into healthy and unhealthy groups, assisting both the vendor and customer in obtaining the right quality for their requirements. The primary cause of the low production of tea in various parts of the country is the late detection of diseased leaves. Literature suggests that the current AI techniques can detect but have a high level of similarities. Hence, the objective is to identify diseased leaf images at early stage. Methods: Earlier studies have used mostly CNN for classification which is more time consuming, and they create their own parameters. In this study, we extracted the textual features of images by using GLCM, and further classification was implemented with the help of SMOTE and a soft voting classifier. Findings: The proposed method resulted in optimal timeline without using CNN for classification. The accuracy is better when compared to current methods. Novelty: This work presents a novel approach in calculating textual features of leaves that can be further used for classification or tea-leaf grading purposes. Keywords: Image processing, Feature extraction, GLCM, SMOTE, Voting classifier

Skin Cancer Classification Using Fine-Tuned Transfer Learning of DENSENET-121

Skin cancer diagnosis greatly benefits from advanced machine learning techniques, particularly fine-tuned deep learning models. In our research, we explored the impact of traditional machine learning and fine-tuned deep learning approaches on prediction accuracy. Our findings reveal significant improvements in predictability and accuracy with fine-tuning, particularly evident in deep learning models. The CNN, SVM, and Random Forest Classifier achieved high accuracy. However, fine-tuned deep learning models such as EfficientNetB0, ResNet34, VGG16, Inception _v3, and DenseNet121 demonstrated superior performance. To ensure comparability, we fine-tuned these models by incorporating additional layers, including one flatten layer and three densely interconnected layers. These layers play a crucial role in enhancing model efficiency and performance. The flatten layer preprocesses multidimensional feature maps, facilitating efficient information flow, while subsequent dense layers refine feature representations, capturing intricate patterns and relationships within the data. Leveraging LeakyReLU activation functions in the dense layers mitigates the vanishing gradient problem and promotes stable training. Finally, the output dense layer with a sigmoid activation function simplifies decision making for healthcare professionals by providing binary classification output. Our study underscores the significance of incorporating additional layers in fine-tuned neural network models for skin cancer classification, offering improved accuracy and reliability in diagnosis.

SwinDFU-Net: Deep learning transformer network for infection identification in diabetic foot ulcer.

The identification of infection in diabetic foot ulcers (DFUs) is challenging due to variability within classes, visual similarity between classes, reduced contrast with healthy skin, and presence of artifacts. Existing studies focus on visual characteristics and tissue classification rather than infection detection, critical for assessing DFUs and predicting amputation risk. To address these challenges, this study proposes a deep learning model using a hybrid CNN and Swin Transformer architecture for infection classification in DFU images. The aim is to leverage end-to-end mapping without prior knowledge, integrating local and global feature extraction to improve detection accuracy. The proposed model utilizes a hybrid CNN and Swin Transformer architecture. It employs the Grad CAM technique to visualize the decision-making process of the CNN and Transformer blocks. The DFUC Challenge dataset is used for training and evaluation, emphasizing the model's ability to accurately classify DFU images into infected and non-infected categories. The model achieves high performance metrics: sensitivity (95.98%), specificity (97.08%), accuracy (96.52%), and Matthews Correlation Coefficient (0.93). These results indicate the model's effectiveness in quickly diagnosing DFU infections, highlighting its potential as a valuable tool for medical professionals. The hybrid CNN and Swin Transformer architecture effectively combines strengths from both models, enabling accurate classification of DFU images as infected or non-infected, even in complex scenarios. The use of Grad CAM provides insights into the model's decision process, aiding in identifying infected regions within DFU images. This approach shows promise for enhancing clinical assessment and management of DFU infections.

Research on Default Classification of Unbalanced Credit Data Based on PixelCNN-WGAN

Personal credit assessment plays a crucial role in the financial system, which not only relates to the financial activities of individuals but also affects the overall credit system and economic health of society. However, the current problem of data imbalance affecting classification results in the field of personal credit assessment has not been fully solved. In order to solve this problem better, we propose a data-enhanced classification algorithm based on a Pixel Convolutional Neural Network (PixelCNN) and a Generative Adversarial Network (Wasserstein GAN, WGAN). Firstly, the historical data containing borrowers’ borrowing information are transformed into grayscale maps; then, data enhancement of default images is performed using the improved PixelCNN-WGAN model; and finally, the expanded image dataset is inputted into the CNN, AlexNet, SqueezeNet, and MobileNetV2 for classification. The results on the real dataset LendingClub show that the data enhancement algorithm designed in this paper improves the accuracy of the four algorithms by 1.548–3.568% compared with the original dataset, which can effectively improve the classification effect of the credit data, and to a certain extent, it provides a new idea for the classification task in the field of personal credit assessment.

Corrosion image classification method based on EfficientNetV2

Corrosion is one of the key factors leading to material failure, which can occur in facilities and equipment closely related to people's lives, causing structural damage and thus affecting the safety of people's lives and property. To identify corrosion more effectively across multiple facilities and equipment, this paper utilizes a corrosion binary classification dataset containing various materials to develop a CNN classification model for better detection and distinction of material corrosion, using a methodological paradigm of transfer learning and fine-tuning. The proposed model implementation initially uses data augmentation to enhance the dataset and employs different sizes of EfficientNetV2 for training, evaluated using Confusion Matrix, ROC curve, and the values of Precision, Recall, and F1-score. To further enhance the testing results, this paper focuses on the impact of using the Global Average Pooling layer versus the Global Max Pooling layer, as well as the number of fine-tuning layers. The results show that the Global Average Pooling layer performs better, and EfficientNetV2B0 with a fine-tuning rate of 20 %, and EfficientNetV2S with a fine-tuning rate of 15 %, achieve the highest testing accuracy of 0.9176, an ROC-AUC value of 0.97, and Precision, Recall, and F1-Score values exceeding 0.9. These findings can be served as a reference for other corrosion classification models which uses EfficientNetV2.

Explaining Deep Learning Models Applied in Histopathology: Current Developments and the Path to Sustainability.

The digital pathology landscape is in continuous expansion. The digitalization of slides using WSIs (Whole Slide Images) fueled the capacity of automatic support for diagnostics. The paper presents an overview of the current state of the art methods used in histopathological practice for explaining CNN classification useful for histopathological experts. Following the study we observed that histopathological deep learning models are still underused and that the pathologists do not trust them. Also we need to point out that in order to get a sustainable use of deep learning we need to get the experts to trust the models. In order to do that, they need to understand how the results are generated and how this information correlates with their prior knowledge and for obtaining this they can use the methods highlighted in this study.

Utilizing machine learning to analyze trunk movement patterns in women with postpartum low back pain

This paper presents an analysis of trunk movement in women with postnatal low back pain using machine learning techniques. The study aims to identify the most important features related to low back pain and to develop accurate models for predicting low back pain. Machine learning approaches showed promise for analyzing biomechanical factors related to postnatal low back pain (LBP). This study applied regression and classification algorithms to the trunk movement proposed dataset from 100 postpartum women, 50 with LBP and 50 without. The Optimized optuna Regressor achieved the best regression performance with a mean squared error (MSE) of 0.000273, mean absolute error (MAE) of 0.0039, and R2 score of 0.9968. In classification, the Basic CNN and Random Forest Classifier both attained near-perfect accuracy of 1.0, the area under the receiver operating characteristic curve (AUC) of 1.0, precision of 1.0, recall of 1.0, and F1-score of 1.0, outperforming other models. Key predictive features included pain (correlation of -0.732 with flexion range of motion), range of motion measures (flexion and extension correlation of 0.662), and average movements (correlation of 0.957 with flexion). Feature selection consistently identified pain, flexion, extension, lateral flexion, and average movement as influential across methods. While limited to this initial dataset and constrained by generalizability, machine learning offered quantitative insight. Models accurately regressed (MSE < 0.01, R2 > 0.95) and classified (accuracy > 0.94) trunk biomechanics distinguishing LBP. Incorporating additional demographic, clinical, and patient-reported factors may enhance individualized risk prediction and treatment personalization. This preliminary application of advanced analytics supported machine learning's potential utility for both LBP risk determination and outcome improvement. This study provides valuable insights into the use of machine learning techniques for analyzing trunk movement in women with postnatal low back pain and can potentially inform the development of more effective treatments.Trial registration: The trial was designed as an observational and cross-section study. The study was approved by the Ethical Committee in Deraya University, Faculty of Pharmacy, (No: 10/2023). According to the ethical standards of the Declaration of Helsinki. This study complies with the principles of human research. Each patient signed a written consent form after being given a thorough description of the trial. The study was conducted at the outpatient clinic from February 2023 till June 30, 2023.

Novel Deep CNNs Explore Regions, Boundaries, and Residual Learning for COVID-19 Infection Analysis in Lung CT.

COVID-19 poses a global health crisis, necessitating precise diagnostic methods for timely containment. However, accurately delineating COVID-19-affected regions in lung CT scans is challenging due to contrast variations and significant texture diversity. In this regard, this study introduces a novel two-stage classification and segmentation CNN approach for COVID-19 lung radiological pattern analysis. A novel Residual-BRNet is developed to integrate boundary and regional operations with residual learning, capturing key COVID-19 radiological homogeneous regions, texture variations, and structural contrast patterns in the classification stage. Subsequently, infectious CT images undergo lesion segmentation using the newly proposed RESeg segmentation CNN in the second stage. The RESeg leverages both average and max-pooling implementations to simultaneously learn region homogeneity and boundary-related patterns. Furthermore, novel pixel attention (PA) blocks are integrated into RESeg to effectively address mildly COVID-19-infected regions. The evaluation of the proposed Residual-BRNet CNN in the classification stage demonstrates promising performance metrics, achieving an accuracy of 97.97%, F1-score of 98.01%, sensitivity of 98.42%, and MCC of 96.81%. Meanwhile, PA-RESeg in the segmentation phase achieves an optimal segmentation performance with an IoU score of 98.43% and a dice similarity score of 95.96% of the lesion region. The framework's effectiveness in detecting and segmenting COVID-19 lesions highlights its potential for clinical applications.

Deep Learning-Based Hand-Drawn Illustration in Packaging Design of Cultural and Creative Products

Packaging design is a critical component of product marketing and branding, encompassing the visual and structural elements that encase and present goods to consumers. The hand-drawn illustration is a timeless art form that embodies the unique style, skill, and creativity of the artist's hand. This paper presents a novel approach to deep learning techniques for enhancing packaging design through the classification of hand-drawn illustrations. The proposed model is stated as a Weighted Augmented Deep Generative Network (WADGN). The proposed WADGN model uses the augmentation network for the generation of the augmented images for the creative products. With the augmented images features are extracted in the hand-drawn illustration of the products. The extracted features are implemented with the weighted augmented feature vector for the application of the generative deep learning network. The proposed WADGN model uses the feature vector of the deep learning model for the design of creative product design. With the deep learning the creative features of the hand-drawn illustration are classified for the creative package design. Simulation results demonstrated that proposed WADGN model higher performance than the conventional technique such as CNN, LSTM and SVM classifier. The proposed WADGN model achieves the ~21% higher performance than the SVM, ~16% than the LSTM and ~9% improvement than the CNN model.