Intra-class Similarity Research Articles

Underwater Incomplete Target Recognition Network via Generating Feature Module

A complex and changeable underwater archaeological environment leads to the lack of target features in the collected images, affecting the accuracy of target detection. Meanwhile, the difficulty in obtaining underwater archaeological images leads to less training data, resulting in poor generalization performance of the recognition algorithm. For these practical issues, we propose an underwater incomplete target recognition network via generating feature module (UITRNet). Specifically, for targets that lack features, features are generated by dual discriminators and generators to improve target detection accuracy. Then, multilayer features are fused to extract regions of interest. Finally, supervised contrastive learning is introduced into few-shot learning to improve the intraclass similarity and interclass distance of the target and enhance the generalization of the algorithm. The UIFI dataset is produced to verify the effectiveness of the algorithm in this paper. The experimental results show that the mean average precision (mAP) of our algorithm was improved by 0.86% and 1.29% under insufficient light and semiburied interference, respectively. The mAP for ship identification reached the highest level under all four sets of experiments.

Discriminative Fisher Embedding Dictionary Transfer Learning for Object Recognition.

In transfer learning model, the source domain samples and target domain samples usually share the same class labels but have different distributions. In general, the existing transfer learning algorithms ignore the interclass differences and intraclass similarities across domains. To address these problems, this article proposes a transfer learning algorithm based on discriminative Fisher embedding and adaptive maximum mean discrepancy (AMMD) constraints, called discriminative Fisher embedding dictionary transfer learning (DFEDTL). First, combining the label information of source domain and part of target domain, we construct the discriminative Fisher embedding model to preserve the interclass differences and intraclass similarities of training samples in transfer learning. Second, an AMMD model is constructed using atoms and profiles, which can adaptively minimize the distribution differences between source domain and target domain. The proposed method has three advantages: 1) using the Fisher criterion, we construct the discriminative Fisher embedding model between source domain samples and target domain samples, which encourages the samples from the same class to have similar coding coefficients; 2) instead of using the training samples to design the maximum mean discrepancy (MMD), we construct the AMMD model based on the relationship between the dictionary atoms and profiles; thus, the source domain samples can be adaptive to the target domain samples; and 3) the dictionary learning is based on the combination of source and target samples which can avoid the classification error caused by the difference among samples and reduce the tedious and expensive data annotation. A large number of experiments on five public image classification datasets show that the proposed method obtains better classification performance than some state-of-the-art dictionary and transfer learning methods. The code has been available at https://github.com/shilinrui/DFEDTL.

ICL-Net: Global and Local Inter-Pixel Correlations Learning Network for Skin Lesion Segmentation.

Skin lesion segmentation is a fundamental procedure in computer-aided melanoma diagnosis. However, due to the diverse shape, variable size, blurry boundary, and noise interference of lesion regions, existing methods may struggle with the challenge of inconsistency within classes and indiscrimination between classes. In view of this, we propose a novel method to learn and model inter-pixel correlations from both global and local aspects, which can increase inter-class variances and intra-class similarities. Specifically, under the encoder-decoder architecture, we first design a pyramid transformer inter-pixel correlations (PTIC) module, aiming at capturing the non-local context information of different levels and further exploring the global pixel-level relationship to deal with the large variance of shape and size. Further, we devise a local neighborhood metric learning (LNML) module to strengthen the local semantic correlations learning capability and increase the separability between classes in the feature space. These two modules can complementarily strengthen the feature representation capability via exploiting the inter-pixel semantic correlations, thus further improving intra-class consistency and inter-class variance. Comprehensive experiments are performed on public skin lesion segmentation datasets: ISIC 2018, ISIC2016, and PH2, and experimental results demonstrate that the proposed method achieves better segmentation performance than other state-of-the-art methods.

Spectral–Spatial Masked Transformer With Supervised and Contrastive Learning for Hyperspectral Image Classification

Recently, due to the powerful capability at modeling the long-range relationships, Transformer-based methods have been widely explored in many research areas including hyperspectral image (HSI) classification. However, because of lots of trainable parameters and the lack of inductive bias, it is difficult to train a Transformer-based HSI classifier, especially when the number of training samples is limited. To address this issue, in this study, spectral-spatial masked Transformer (SS-MTr) is explored for HSI classification, which uses a two-stage training strategy. In the first stage, SS-MTr pre-trains a vanilla Transformer via reconstruction from masked HSI inputs, which embeds the local inductive bias into the Transformer. In the second stage, the well pre-trained Transformer is cooperated with a fully connected layer and then fine-tuned for the HSI classification. Furthermore, in order to incorporate discriminative feature learning into the SS-MTr, three SS-MTr-based methods, including contrastive SS-MTr (C-SS-MTr), supervised SS-MTr (S-SS-MTr), and supervised contrastive SS-MTr (SC-SS-MTr) are proposed by adding extra branches for specific tasks in parallel with the existing reconstruction task. Specifically, the proposed C-SS-MTr adds a contrastive loss which brings instance discriminability. Besides, the proposed S-SS-MTr builds an extra classification branch for embracing inter-class discriminability and intra-class similarity. Moreover, the proposed SC-SS-MTr combines C-SS-MTr and S-SS-MTr for better generalization. The proposed SS-MTr, C-SS-MTr, S-SS-MTr, and SC-SS-MTr are tested on three popular hyperspectral datasets (i.e., Indian Pines, Pavia University, and Houston). The obtained results reveal that the proposed models achieve competitive results compared with the state-of-the-art HSI classification methods. Code is available at https://github.com/mengduanjinghua/SS-MTr.

Semantic-Aware Region Loss for Land-Cover Classification

Integrating superpixel segmentation into convolutional neural networks is known to be effective in enhancing the accuracy of land-cover classification. However, most of existing methods accomplish such integration by focusing on the development of new network architectures, which suffer from several flaws: 1) conflicts between general superpixels and semantic labels introduce noise into the training, especially at object boundaries; 2) absence of training guidance for superpixels leads to ineffective regional feature learning; 3) unnecessary superpixel segmentation in the testing stage not only increases the computational burden but also incurs jagged edges. In this study, we propose a novel semantic-aware region (SARI) loss to guide the effective learning of regional features with superpixels for accurate land-cover classification. The key idea of the proposed method is to reduce the feature variance inside and between homogeneous superpixels while enlarging feature discrepancy between heterogeneous ones. The SARI loss is thus designed with three sub-parts, including superpixel variance loss, intra-class similarity loss and inter-class distance loss. We also develop semantic superpixels to assist in the network training with SARI loss while overcoming the limitations of general superpixels. Extensive experiments on two challenging datasets demonstrate that the SARI loss can facilitate regional feature learning, achieving state-of-the-art performance with mIoU scores of around 97.11% and 73.99% on GID and DeepGlobe datasets, respectively. Code is available at: <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/geovsion/SARI</uri> .

Fine-Grained Object Detection in Remote Sensing Images via Adaptive Label Assignment and Refined-Balanced Feature Pyramid Network

Object detection in high-resolution remote sensing images remains a challenging task due to the uniqueness of its viewing perspective, complex background, arbitrary orientation, etc. For fine-grained object detection in high-resolution remote sensing images, the high intra-class similarity is even more severe, which makes it difficult for the object detector to recognize the correct classes. In this article, we propose the refined and balanced feature pyramid network (RB-FPN) and center-scale aware (CSA) label assignment strategy to address the problems of fine-grained object detection in remote sensing images. RB-FPN fuses features from different layers and suppresses background information when focusing on regions that may contain objects, providing high-quality semantic information for fine-grained object detection. Intersection over Union (IoU) is usually applied to select the positive candidate samples for training. However, IoU is sensitive to the angle variation of oriented objects with large aspect ratios, and a fixed IoU threshold will cause the narrow oriented objects without enough positive samples to participate in the training. In order to solve the problem, we propose the CSA label assignment strategy that adaptively adjusts the IoU threshold according to statistical characteristics of oriented objects. Experiments on FAIR1M dataset demonstrate that the proposed approach is superior. Moreover, the proposed method was applied to the fine-grained object detection in high-resolution optical images of 2021 Gaofen challenge. Our team ranked sixth and was awarded as the winning team in the final.

Hand Gesture Recognition for Disabled People Using Bayesian Optimization with Transfer Learning

Sign language recognition can be treated as one of the efficient solutions for disabled people to communicate with others. It helps them to convey the required data by the use of sign language with no issues. The latest developments in computer vision and image processing techniques can be accurately utilized for the sign recognition process by disabled people. American Sign Language (ASL) detection was challenging because of the enhancing intraclass similarity and higher complexity. This article develops a new Bayesian Optimization with Deep Learning-Driven Hand Gesture Recognition Based Sign Language Communication (BODL-HGRSLC) for Disabled People. The BODL-HGRSLC technique aims to recognize the hand gestures for disabled people’s communication. The presented BODL-HGRSLC technique integrates the concepts of computer vision (CV) and DL models. In the presented BODL-HGRSLC technique, a deep convolutional neural network-based residual network (ResNet) model is applied for feature extraction. Besides, the presented BODL-HGRSLC model uses Bayesian optimization for the hyperparameter tuning process. At last, a bidirectional gated recurrent unit (BiGRU) model is exploited for the HGR procedure. A wide range of experiments was conducted to demonstrate the enhanced performance of the presented BODL-HGRSLC model. The comprehensive comparison study reported the improvements of the BODL-HGRSLC model over other DL models with maximum accuracy of 99.75%.

Multiscale Superpixelwise Prophet Model for Noise-Robust Feature Extraction in Hyperspectral Images

Despite of various approaches proposed to smooth the hyperspectral images (HSIs) before feature extraction, the efficacy is still affected by the noise, even using the corrected dataset with the noisy and water absorption bands discarded. In this study, a novel spectral-spatial feature mining framework, Multiscale Superpixelwise Prophet Model (MSPM), is proposed for noise-robust feature extraction and effective classification of the HSI. The prophet model is highly noise-robust for deeply digging into the complex structured features thus enlarging interclass diversity and improving intraclass similarity. First, the superpixelwise segmentation is produced from the first three principal components of an HSI to group pixels into regions with adaptively determined sizes and shapes. A multiscale prophet model is utilized to extract the multiscale informative trend components from the average spectrum of each superpixel. Taking the multiscale trend signal as the input feature, the HSI data are classified superpixelwisely, which is further refined by a majority vote based decision fusion. Comprehensive experiments on three publicly available datasets have fully validated the efficacy and robustness of our MSPM model when benchmarked with eleven state-of-the-art algorithms, including six spectral-spatial methods and five deep learning ones. Besides, MSPM also shows superiority under limited training samples, due to the combined strategies of superpixelwise fusion and multiscale fusion. Our model has provided a useful solution for noise-robust feature extraction as it achieves superior HSI classification even from the uncorrected dataset without prefiltering the water absorption and noisy bands.

Zero-Shot SAR Target Recognition Based on Classification Assistance

As one of main active learning methods, zero-shot target recognition with synthetic aperture radar (SAR) data has received considerable attention in recent years. Its goal is to distinguish new targets from the known classes without requiring additional training data. Existing zero-shot learning (ZSL) methods perform well on optical targets, but they fail to recognize zero-shot SAR targets. Due to the strong similarity of SAR targets between different classes, the ZSL task usually suffers from the distribution concentration problem. To tackle this problem, a Dual Branch Auto-Encoder (DBAE) network is proposed in this letter. DBAE effectively alleviates the distribution concentration problem by adding a classification assistance net. Its dual branch structure is specially designed for further improving the intra-class similarity and inter-class dissimilarity of SAR targets in the embedding space. By training with the defined hybrid loss function, DBAE automatically builds a stable embedding space. Extensive experiments on the public data set of Moving and Stationary Target Acquisition and Recognition (MSTAR) show that DBAE is of rational design and provides better or comparable ZSL recognition results.

Estimates of the Complexity of Detecting Types of DDOS Attacks

The problem of substantiating decisions made in the field of information security through estimates of the complexity of detecting types of DDOS attacks is considered. Estimates are a quantitative measure of a particular type of attack relative to normal network operation traffic data in its own feature space. Own space is represented by a set of informative features. To assess the complexity of detecting types of DDOS attacks, a measure of compactness by latent features on the numerical axis was used. The values of this measure were calculated as the product of intraclass similarity and interclass difference. It is shown that compactness in terms of latent features in its own space is higher than in the entire space. The values of latent features were calculated using the method of generalized estimates. According to this method, objects of normal traffic and a specific type of attack are considered as opposition to each other. An informative feature set is the result of an algorithm that uses the rules of hierarchical agglomerative grouping. At the first step, the feature with the maximum weight value is included in the set. The grouping rules apply the feature invariance property to the scales of their measurements. An analysis of the complexity of detection for 12 types of DDOS attacks is given. The power of sets of informative features ranged from 3 to 16.

An open-set fault diagnosis framework for MMCs based on optimized temporal convolutional network

Reliability of the modular multilevel converters (MMCs) provides a vital guarantee for the uninterrupted operation of a system. Insulated Gate Bipolar Transistors (IGBTs) open circuit fault diagnosis is a common challenge in MMCs applications. In this paper, a novel open-set fault diagnosis framework called Multiscale-AAM-OTCN is proposed to solve both the known and unknown fault diagnosis problems of MMCs by outputting current signals. First, batch normalization and layer normalization are introduced into the original Temporal Convolutional Network (TCN) model to accelerate convergence and promote the generalization ability of the model for different tasks. Second, to strengthen the feature extraction ability of the model, the multiscale coordinate residual attention (MCRA) mechanism is designed for the one-dimensional (1D) current signal to improve the performance and stability of the method. Compared with recently developed attention mechanisms such as convolutional block attention module (CBAM), efficient channel attention (ECA), simple, parameter-free attention module (SimAM) and coordinate attention (CA), the proposed MCRA exhibits better performance in MMCs fault diagnosis tasks. Finally, the additive angular margin (AAM) loss and local outlier factor (LOF) algorithm are integrated into the Multiscale-OTCN framework to identify the density difference between known and unknown fault clusters by controlling the intra-class similarity and inter-class variance of the samples. The experimental results demonstrate the feasibility of the proposed fault diagnosis framework for known and unknown fault diagnoses.

Improved deep metric learning with local neighborhood component analysis

Deep metric learning aims to learn a discriminative feature space in which features have larger intra-class similarities and smaller inter-class similarities. Most recent studies mainly focus on designing different loss functions or sampling strategies, while ignoring a crucial limitation caused by mini-batch training. We argue that existing mini-batch-based approaches do not explore the global structure similarities among samples in feature space. As a result, instances and their k-nearest neighbors may not be semantically consistent. To this end, we propose a method, dubbed Local Neighborhood Component Analysis (LNCA), to improve deep metric learning. Specifically, LNCA leverages a feature memory bank, storing the feature vectors of all instances, to estimate the global structure similarities and determine the k nearest neighbors of samples in the feature space. Further, in order to refine the local neighborhood components of samples, LNCA introduces a metric to attract the positive neighbors and repulse the negative neighbors simultaneously. LNCA is a plug-and-play module and can be integrated into a general DML framework. Experimental results show that it can boost the generalization performance of existing DML approaches significantly.

Weather Forecast Based on Color Cloud Image Recognition under the Combination of Local Image Descriptor and Histogram Selection

Numerous researchers have used machine vision in recent years to identify and categorize clouds according to their volume, shape, thickness, height, and coverage. Due to the significant variations in illumination, climate, and distortion that frequently characterize cloud images as a type of naturally striated structure, the Local Binary Patterns (LBP) descriptor and its variants have been proposed as feature extraction methods for characterizing natural texture images. Rotation invariance, low processing complexity, and resistance to monotonous brightness variations are characteristics of LBP. The disadvantage of LBP is that it produces binary data that are extremely noise-sensitive and it struggles on regions of the image that are “flat” because it depends on intensity differences. This paper considers the Local Ternary Patterns (LTP) feature to overcome the drawbacks of the LBP feature. We also propose the fusion of color characteristics, LBP features, and LTP features for the classification of cloud/sky images. Morover, this study proposes to apply the Intra-Class Similarity (ICS) technique, a histogram selection approach, with the goal of minimizing the number of histograms for characterizing images. The proposed approach achieves better performance of recognition with less features in use by fusing LBP and LTP features and using the ICS technique to choose potential histograms.

LDAformer: predicting lncRNA-disease associations based on topological feature extraction and Transformer encoder.

The identification of long noncoding RNA (lncRNA)-disease associations is of great value for disease diagnosis and treatment, and it is now commonly used to predict potential lncRNA-disease associations with computational methods. However, the existing methods do not sufficiently extract key features during data processing, and the learning model parts are either less powerful or overly complex. Therefore, there is still potential to achieve better predictive performance by improving these two aspects. In this work, we propose a novel lncRNA-disease association prediction method LDAformer based on topological feature extraction and Transformer encoder. We construct the heterogeneous network by integrating the associations between lncRNAs, diseases and micro RNAs (miRNAs). Intra-class similarities and inter-class associations are presented as the lncRNA-disease-miRNA weighted adjacency matrix to unify semantics. Next, we design a topological feature extraction process to further obtain multi-hop topological pathway features latent in the adjacency matrix. Finally, to capture the interdependencies between heterogeneous pathways, a Transformer encoder based on the global self-attention mechanism is employed to predict lncRNA-disease associations. The efficient feature extraction and the intuitive and powerful learning model lead to ideal performance. The results of computational experiments on two datasets show that our method outperforms the state-of-the-art baseline methods. Additionally, case studies further indicate its capability to discover new associations accurately.

Geometry-Aware Graph Embedding Projection Metric Learning for Image Set Classification

By describing image sets as linear subspaces on the Grassmann manifold, image set classification has received persistent attention. Despite the success made so far, the unhelpfully intraclass diversity and interclass similarity remain two key challenges in finding an effective lower dimensional feature space for similarity measurement. To explore a feasible solution to these issues, we propose a geometry-aware graph embedding projection metric learning (GEPML) algorithm. The proposed approach first constructs the interclass and the intraclass similarity graphs on the Grassmann manifold, aiming to exploit the local structural information of the data manifold. Besides, we generalize the Euclidean collaborative representation mechanism to the Grassmann manifold to adaptively perform graph learning. Then, to learn the embedding mapping and the similarity metric jointly, we formulate the Grassmannian dimensionality reduction (GDR) problem into an elaborately designed metric learning regularization term. The proposed algorithm is appraised on five benchmarking data sets and the competitive experimental results demonstrate its feasibility and effectiveness.

Atom Search Optimization with Deep Learning Enabled Arabic Sign Language Recognition for Speaking and Hearing Disability Persons.

Sign language has played a crucial role in the lives of impaired people having hearing and speaking disabilities. They can send messages via hand gesture movement. Arabic Sign Language (ASL) recognition is a very difficult task because of its high complexity and the increasing intraclass similarity. Sign language may be utilized for the communication of sentences, letters, or words using diverse signs of the hands. Such communication helps to bridge the communication gap between people with hearing impairment and other people and also makes it easy for people with hearing impairment to express their opinions. Recently, a large number of studies have been ongoing in developing a system that is capable of classifying signs of dissimilar sign languages into the given class. Therefore, this study designs an atom search optimization with a deep convolutional autoencoder-enabled sign language recognition (ASODCAE-SLR) model for speaking and hearing disabled persons. The presented ASODCAE-SLR technique mainly aims to assist the communication of speaking and hearing disabled persons via the SLR process. To accomplish this, the ASODCAE-SLR technique initially pre-processes the input frames by a weighted average filtering approach. In addition, the ASODCAE-SLR technique employs a capsule network (CapsNet) feature extractor to produce a collection of feature vectors. For the recognition of sign language, the DCAE model is exploited in the study. At the final stage, the ASO algorithm is utilized as a hyperparameter optimizer which in turn increases the efficacy of the DCAE model. The experimental validation of the ASODCAE-SLR model is tested using the Arabic Sign Language dataset. The simulation analysis exhibit the enhanced performance of the ASODCAE-SLR model compared to existing models.

Deep multi-scale resemblance network for the sub-class differentiation of adrenal masses on computed tomography images

ObjectiveThe accurate classification of mass lesions in the adrenal glands (‘adrenal masses’), detected with computed tomography (CT), is important for diagnosis and patient management. Adrenal masses can be benign or malignant and benign masses have varying prevalence. Classification methods based on convolutional neural networks (CNNs) are the state-of-the-art in maximizing inter-class differences in large medical imaging training datasets. The application of CNNs, to adrenal masses is challenging due to large intra-class variations, large inter-class similarities and imbalanced training data due to the size of the mass lesions. MethodsWe developed a deep multi-scale resemblance network (DMRN) to overcome these limitations and leveraged paired CNNs to evaluate the intra-class similarities. We used multi-scale feature embedding to improve the inter-class separability by iteratively combining complementary information produced at different scales of the input to create structured feature descriptors. We augmented the training data with randomly sampled paired adrenal masses to reduce the influence of imbalanced training data. ResultsWe used 229 CT scans of patients with adrenal masses for evaluation. In a five-fold cross-validation, our method had the best results (89.52 % in accuracy) when compared to the state-of-the-art methods (p < 0.05). We conducted a generalizability analysis of our method on the ImageCLEF 2016 competition dataset for medical subfigure classification, which consists of a training set of 6776 images and a test set of 4166 images across 30 classes. Our method achieved better classification performance (85.90 % in accuracy) when compared to the existing methods and was competitive when compared with methods that require additional training data (1.47 % lower in accuracy). ConclusionOur DMRN sub-classified adrenal masses on CT and was superior to state-of-the-art approaches.

Twin relaxed least squares regression with classwise mean constraint for image classification

• A novel twin relaxed regression model is introduced for image classification. • A relaxed target matrix together with a twin matrix provide more degrees of freedom to fit the class labels. • Enlarged interclass margins for improved classification. • Adaptively maximize the intraclass similarity with a classwise mean constraint. • Experimental results show that the proposed method outperforms state-of-the-art algorithms in terms of classification rate. This paper presents a twin relaxed least squares regression (TRLSR) framework with classwise mean constraint for image classification. The primary objective of TRLSR is to learn discriminative projections with enhanced interclass margins while preserving the intrinsic structure of the data. To this end, we introduce a relaxed regression target matrix together with a twin matrix to allow greater flexibility in learning the projections compared to using the conventional strict binary label matrix. In addition, a classwise mean constraint is introduced to retain the intraclass similarity of the data, which is beneficial in learning more discriminative projections. An ℓ 2,1 -norm based regularization on the optimized projections is incorporated to extract more significant features while limiting the impact of noise and overfitting. The performance of the proposed technique on several public data sets for face recognition, object classification, action recognition and scene classification applications is demonstrated. The proposed method is shown to outperform the state-of-the-art approaches.

Center Prediction Loss for Re-identification

The training loss function that enforces certain training sample distribution patterns plays a critical role in building a re-identification (ReID) system. Besides the basic requirement of discrimination, i.e., the features corresponding to different identities should not be mixed, additional intra-class distribution constraints, such as features from the same identities should be close to their centers, have been adopted to construct losses. Despite the advances of various new loss functions, it is still challenging to strike the balance between the need of reducing the intra-class variation and allowing certain distribution freedom. Traditional intra-class losses try to shrink samples of the same class into one point in the feature space and may easily drop their intra-class similarity structure. In this paper, we propose a new loss based on center predictivity, that is, a sample must be positioned in a location of the feature space such that from it we can roughly predict the location of the center of same-class samples. The prediction error is then regarded as a loss called Center Prediction Loss (CPL). Unlike most existing metric learning loss functions, CPL involves learnable parameters, i.e., the center predictor, which brings a remarkable change in the properties of the loss. In particular, it allows higher freedom in intra-class distributions. And the parameters in CPL will be discarded after training. Extensive experiments on various real-world ReID datasets show that the proposed loss can achieve superior performance and can also be complementary to existing losses.

An Extra-Contrast Affinity Network for Facial Expression Recognition in the Wild

Learning discriminative features for facial expression recognition (FER) in the wild is a challenging task due to the significant intra-class variations, inter-class similarities, and extreme class imbalances. In order to solve these issues, a contrastive-learning-based extra-contrast affinity network (ECAN) method is proposed. The ECAN consists of a feature processing network and two proposed loss functions, namely extra negative supervised contrastive loss (ENSC loss) and multi-view affinity loss (MVA loss). The feature processing network provides current and historical deep features to satisfy the necessary conditions for these loss functions. Specifically, the ENSC loss function simultaneously considers many positive samples and extra negative samples from other minibatches to maximize intra-class similarity and the inter-class separation of deep features, while also automatically turning the attention of the model to majority and minority classes to alleviate the class imbalance issue. The MVA loss function improves upon the center loss function by leveraging additional deep feature groups from other minibatches to dynamically learn more accurate class centers and further enhance the intra-class compactness of deep features. The numerical results obtained using two public wild FER datasets (RAFDB and FER2013) indicate that the proposed method outperforms most state-of-the-art models in FER.