Intra-class Variations Research Articles

CASF-MNet: multi-scale network with cross attention mechanism and spatial dimension feature fusion for maize leaf disease detection

Maize is an important food crop, and accurate identification of maize leaf diseases is significant for disease control. Aiming at the complexity of the color and texture features of diseases with inter-class similarity and intra-class variability, a two-channel multi-scale network model with a cross-attention mechanism and spatial dimension feature fusion (CASF-MNet) was designed. Firstly, aiming at the complexity of color features, the HSV color Sub-network (CSNet) was designed to extract the color features of disease spots. Then, aiming at the complexity of texture features, Texture Sub-Network (TSNet) and Cross Attention mechanism (CAM) were proposed to enhance the texture features of disease spots. Finally, to better fuse color and texture features, a cross-spatial feature fusion mechanism (FFM) was proposed. In the five-fold cross-validation of the self-built dataset, the average F1-Score and average Accuracy of CASF-MNet reach 97.17% and 97.02%. In the test experiments of two public datasets, the F1-Score of CASF-MNet reaches 98.13% and 98.71%, and the Accuracy reaches 97.44% and 98.67%, which are better than the seven most advanced leaf disease identification methods. It is proved that CASF-MNet is advanced and stable, and can assist disease detection personnel in maize disease detection in field environments.

Tuberculin skin test result detection method based on CSN-II and improved OTSU method

The tuberculin skin test (TST) is an intradermal test used to diagnose Type IV hypersensitivity reactions caused by Mycobacterium tuberculosis infection. Segmentation of TST result images provides a foundation for large-scale tuberculosis screening and auxiliary diagnosis. This paper presents a specialized method for identifying TST results. Initially, a clustering approach is employed to reduce pixel complexity, followed by a linear transformation using CSN-II to enhance the original RGB space with robust color space properties. Subsequently, high-probability pixel points are located, and their Gaussian kernel convolution range is determined using the Bhattacharyya Distance. Through convolution and iterative feature amplification, the target characteristics are progressively enhanced. Finally, an improved OTSU method is proposed for segmenting TST result images. In this method, an adaptive entropy threshold is utilized to reduce the search range of the OTSU method, enhancing the relative contrast between the target and the background. Moreover, a weighted adjustment is applied to the obtained OTSU threshold to prevent drift towards backgrounds with larger intra-class variances. Experimental results demonstrate that the proposed method achieves higher segmentation accuracy and robustness in TST result image segmentation compared to traditional OTSU methods and other improved approaches, such as the neighborhood valley-emphasis method, logarithmic OTSU, and weighted OTSU, Finally, we calculate a relative value is calculated by dividing the remaining number of segmented pixels by the total number of pixels, we then classify the results based on the relative value and in reference to medical diagnostic standards our method is intended to establish an algorithmic basis for rapid screening and classification of tuberculosis on a large scale.

A fish appetite assessment method based on improved ByteTrack and spatiotemporal graph convolutional network

The appetite of fish significantly influences aquaculture efficiency and fish welfare. However, accurately assessing fish appetite has posed a challenging problem. Currently, the study of fish feeding behaviour relies primarily on the overall information obtained from images of fish schools, often overlooking the distinctive behavioural traits of individual fish. Analysing the behaviour of individual fish is hindered by challenges such as intraclass variation and cross-occlusion within real aquaculture environments. To address these challenges, this paper introduces a novel method for assessing appetite based on individual fish behaviour. The ByteTrack model was improved to enable stable tracking of each fish within a school under complex conditions. Additionally, this paper employs the spatiotemporal graph convolutional neural network (ST-GCN) to extract the movement characteristics of individual fish, facilitating accurate appetite assessment. The experimental results demonstrate that the proposed method achieves 98.47% accuracy in appetite assessment, surpassing the performance of other state-of-the-art methods. This paper provides a new opportunity and effective means for analysing fish behaviour and appetite in intricate environments.

Spatial–Temporal Approach and Dataset for Enhancing Cloud Detection in Sentinel-2 Imagery: A Case Study in China

Clouds often cause challenges during the application of optical satellite images. Masking clouds and cloud shadows is a crucial step in the image preprocessing workflow. The absence of a thermal band in products of the Sentinel-2 series complicates cloud detection. Additionally, most existing cloud detection methods provide binary results (cloud or non-cloud), which lack information on thin clouds and cloud shadows. This study attempted to use end-to-end supervised spatial–temporal deep learning (STDL) models to enhance cloud detection in Sentinel-2 imagery for China. To support this workflow, a new dataset for time-series cloud detection featuring high-quality labels for thin clouds and haze was constructed through time-series interpretation. A classification system consisting of six categories was employed to obtain more detailed results and reduce intra-class variance. Considering the balance of accuracy and computational efficiency, we constructed four STDL models based on shared-weight convolution modules and different classification modules (dense, long short-term memory (LSTM), bidirectional LSTM (Bi-LSTM), and transformer). The results indicated that spatial and temporal features were crucial for high-quality cloud detection. The STDL models with simple architectures that were trained on our dataset achieved excellent accuracy performance and detailed detection of clouds and cloud shadows, although only four bands with a resolution of 10 m were used. The STDL models that used the Bi-LSTM and that used the transformer as the classifier showed high and close overall accuracies. While the transformer classifier exhibited slightly lower accuracy than that of Bi-LSTM, it offered greater computational efficiency. Comparative experiments also demonstrated that the usable data labels and cloud detection results obtained with our workflow outperformed the results of the existing s2cloudless, MAJA, and CS+ methods.

TAG-Net: Target Attitude Angle-Guided Network for Ship Detection and Classification in SAR Images

Synthetic aperture radar (SAR) ship detection and classification has gained unprecedented attention due to its important role in maritime transportation. Many deep learning-based detectors and classifiers have been successfully applied and achieved great progress. However, ships in SAR images present discrete and multi-centric features, and their scattering characteristics and edge information are sensitive to variations in target attitude angles (TAAs). These factors pose challenges for existing methods to obtain satisfactory results. To address these challenges, a novel target attitude angle-guided network (TAG-Net) is proposed in this article. The core idea of TAG-Net is to leverage TAA information as guidance and use an adaptive feature-level fusion strategy to dynamically learn more representative features that can handle the target imaging diversity caused by TAA. This is achieved through a TAA-aware feature modulation (TAFM) module. It uses the TAA information and foreground information as prior knowledge and establishes the relationship between the ship scattering characteristics and TAA information. This enables a reduction in the intra-class variability and highlights ship targets. Additionally, considering the different requirements of the detection and classification tasks for the scattering information, we propose a layer-wise attention-based task decoupling detection head (LATD). Unlike general deep learning methods that use shared features for both detection and classification tasks, LATD extracts multi-level features and uses layer attention to achieve feature decoupling and select the most suitable features for each task. Finally, we introduce a novel salient-enhanced feature balance module (SFB) to provide richer semantic information and capture the global context to highlight ships in complex scenes, effectively reducing the impact of background noise. A large-scale ship detection dataset (LSSDD+) is used to verify the effectiveness of TAG-Net, and our method achieves state-of-the-art performance.

A high accuracy method for the sintering condition recognition of rotary kiln

According to the different material sintering conditions, the sintering conditions of alumina rotary kiln can be divided into: super-heated, super-chilled, and normal. In this paper, based on Local Binary Pattern (LBP) and the primary-color method, a novel feature extraction method is proposed to obtain information about temperature in flame images without calibrating camera parameters. Through the analysis of the problem as well as the experimental phenomena, a new classification procedure is devised: in the first step, the super-chilled condition is first separated, in the second step, the normal and super-heat condition are classified. Different feature extraction methods are used in the two steps mentioned above. One is to extract the texture features of pseudo temperature images, and short-time energy is used to describe the dynamic features. The other is to extract the texture features of grey-images by our improved LBPP,Rriu2, and characterize the dynamics with sample entropy and variance. Finally, experimental results show that the our feature extraction method can effectively reduce intra-class variation, increase inter-class variation and receive a high classification accuracy.

Intra-class deep learning object detection on embedded computer system

<span lang="EN-US">Implementation of artificial intelligence tends to be portable, mobile and embeds in embedded computer system (EBD). EBD is a special-purpose computer with limited capacity in a small-form size. Deep learning (DL) had known as cutting edges for object recognition. With DL, object feature extraction analysis is omitted. DL requires large computing resources and capacity. Implement DL algorithm on EBD goal to achieves high detection accuracy and high-efficiency resources. Hence, be able to cope with intra-class variations, and image disturbances. By those challenges and limitations, this study reports the performance of EBD to recognize an object which has high variations in their class, through an optimal raw-input dataset. The raw-input dataset performed optimization process with a supervisor. Yield is the proper optimal input dataset in size. The performance results observed begin from training dataset until evaluation stage of DL. The comparison performs in efficiency resources, loss, validation-loss, timesteps, and detection accuracy by multiclass confusion matrix analysis. This study shows through this purpose method efficient resources are highly archived. Shorter timesteps ensure training stage is successful, and detection accuracy is perfectly archived. In addition, this study proves DL method archived great performances in classifying object that has identical structure.</span>

Edge-Enhanced TempoFuseNet: A Two-Stream Framework for Intelligent Multiclass Video Anomaly Recognition in 5G and IoT Environments

Surveillance video analytics encounters unprecedented challenges in 5G and IoT environments, including complex intra-class variations, short-term and long-term temporal dynamics, and variable video quality. This study introduces Edge-Enhanced TempoFuseNet, a cutting-edge framework that strategically reduces spatial resolution to allow the processing of low-resolution images. A dual upscaling methodology based on bicubic interpolation and an encoder–bank–decoder configuration is used for anomaly classification. The two-stream architecture combines the power of a pre-trained Convolutional Neural Network (CNN) for spatial feature extraction from RGB imagery in the spatial stream, while the temporal stream focuses on learning short-term temporal characteristics, reducing the computational burden of optical flow. To analyze long-term temporal patterns, the extracted features from both streams are combined and routed through a Gated Recurrent Unit (GRU) layer. The proposed framework (TempoFuseNet) outperforms the encoder–bank–decoder model in terms of performance metrics, achieving a multiclass macro average accuracy of 92.28%, an F1-score of 69.29%, and a false positive rate of 4.41%. This study presents a significant advancement in the field of video anomaly recognition and provides a comprehensive solution to the complex challenges posed by real-world surveillance scenarios in the context of 5G and IoT.

Rice seed classification by hyperspectral imaging system: A real-world dataset and a credible algorithm

Rice is one of the three major crops in the world. The hyperspectral identification of rice seeds is of great significance to the development of agriculture and hence has made remarkable achievements. However, the lack of publicly available benchmark datasets with a large number of rice seeds hinders the generalization of models and limits the advancement of the field. To address this issue, this study proposes a publicly accessible dataset for rice seed classification by hyperspectral imaging system. The proposed dataset contains 6 categories of nearly 10,000 seeds each and is characterized by (1) a large scale on the number of seeds per category and the total seed number; (2) a greater adherence to real-world scenarios; and (3) high intra-class diversity and high inter-class similarity. Based on the proposed dataset, an Instance Difficulty-weighted K-Nearest Neighbor algorithm (IDKNN) is further proposed. It is a credible algorithm, emphasizing individual seed accuracy rather than overall performance. Additionally, we explore the effects of different Regions Of Interest (ROI). Experimental results constitute the baseline performance of the proposed dataset. Compared with 10 representative algorithms, the IDKNN algorithm achieves the best performance, especially on the ROI of the entire seed. The dataset can be downloaded from https://github.com/fliery/RSHI60T.

Explicit knowledge transfer of graph-based correlation distillation and diversity data hallucination for few-shot object detection

The performance of few-shot object detection has seen marked improvement through fine-tuning paradigms. However, existing methods often depend on shared parameters to implicitly transfer knowledge without explicit induction. This results in novel-class representations that are easily confused with similar base classes and poorly suited to diverse patterns of variation in the truth distribution. In view of this, the present paper focuses on mining transferable base-class knowledge, which is further subdivided into inter-class correlation and intra-class diversity. First, we design a graph to dynamically capture the relationship between base and novel class representations, and then introduce distillation techniques to tackle the shortage of correlation knowledge in few-shot labels. Furthermore, an efficient diversity knowledge transfer module based on the data hallucination is proposed, which can adaptively disentangle class-independent variation patterns from base-class features and generate additional trainable hallucinated instances for novel classes. Experiments on VOC and COCO datasets confirmed that our proposed method effectively reduces the reliance on novel-class samples and demonstrates superior performance compared to other state-of-the-art baseline methods.

Hyperspectral image classification using Walsh Hadamard transform- based key band selection and deep convolutional neural networks

ABSTRACT Hyperspectral images (HSIs) have transformed the field of remote sensing by providing researchers with a wealth of information about the Earth’s surface. However, analyzing these images can be an overwhelming task due to the presence of overlapping areas, nested regions, and large intra-class variability. Hyperspectral image classification (HSIC) is a crucial part of identifying the various land cover classes present in hyperspectral images. In order to enhance the accuracy of HSIC, researchers utilize the potential of three-dimensional convolutional neural networks (3D-CNNs). With the ability to influence both the spectral and spatial data present in HSIs, 3D-CNNs provide a promising solution to overcome the challenges associated with HSIC. In this paper, a new method for key band selection is proposed to improve the performance of 3D-CNN model. The proposed method selects the most relevant key bands based on Walsh-Hadamard kernel strength features. These key bands are then used to extract overlapping 3D spatial patches, which serve as input to the proposed 3D-CNN model. To evaluate the performance of the 3D-CNN model six standard benchmark datasets are used. The effectiveness of the proposed method improves the performance of 3D-CNN for HSIC.

Research on an Enhanced Multimodal Network for Specific Emitter Identification

Specific emitter identification (SEI) refers to the task of distinguishing similar emitters, especially those of the same type and transmission parameters, which is one of the most critical tasks of electronic warfare. However, SEI is still a challenging task when a feature has low physical representation. Feature representation largely determines the recognition results. Therefore, this article expects to move toward robust feature representation for SEI. Efficient multimodal strategies have great potential for applications using multimodal data and can further improve the performance of SEI. In this research, we introduce a multimodal emitter identification method that explores the application of multimodal data, time-series radar signals, and feature vector data to an enhanced transformer, which employs a conformer block to embed the raw data and integrates an efficient multimodal feature representation module. Moreover, we employ self-knowledge distillation to mitigate overconfident predictions and reduce intra-class variations. Our study reveals that multimodal data provide sufficient information for specific emitter identification. Simultaneously, we propose the CV-CutMixOut method to augment the time-domain signal. Extensive experiments on real radar datasets indicate that the proposed method achieves more accurate identification results and higher feature discriminability.

Learning Semantic Alignment Using Global Features and Multi-Scale Confidence

Semantic alignment aims to establish pixel correspondences between images based on semantic consistency. It can serve as a fundamental component for various downstream computer vision tasks, such as style transfer and exemplar-based colorization, etc. Many existing methods use local features and their cosine similarities to infer semantic alignment. However, they struggle with significant intra-class variation of objects, such as appearance, size, etc. In other words, contents with the same semantics tend to be significantly different in vision. To address this issue, we propose a novel deep neural network of which the core lies in global feature enhancement and adaptive multi-scale inference. Specifically, two modules are proposed: an enhancement transformer for enhancing semantic features with global awareness; a probabilistic correlation module for adaptively fusing multi-scale information based on the learned confidence scores. We use the unified network architecture to achieve two types of semantic alignment, namely, cross-object semantic alignment and cross-domain semantic alignment. Experimental results demonstrate that our method achieves competitive performance on five standard cross-object semantic alignment benchmarks, and outperforms the state of the arts in cross-domain semantic alignment.

Center-Focused Affinity Loss for Class Imbalance Histology Image Classification.

Early-stage cancer diagnosis potentially improves the chances of survival for many cancer patients worldwide. Manual examination of Whole Slide Images (WSIs) is a time-consuming task for analyzing tumor-microenvironment. To overcome this limitation, the conjunction of deep learning with computational pathology has been proposed to assist pathologists in efficiently prognosing the cancerous spread. Nevertheless, the existing deep learning methods are ill-equipped to handle fine-grained histopathology datasets. This is because these models are constrained via conventional softmax loss function, which cannot expose them to learn distinct representational embeddings of the similarly textured WSIs containing an imbalanced data distribution. To address this problem, we propose a novel center-focused affinity loss (CFAL) function that exhibits 1) constructing uniformly distributed class prototypes in the feature space, 2) penalizing difficult samples, 3) minimizing intra-class variations, and 4) placing greater emphasis on learning minority class features. We evaluated the performance of the proposed CFAL loss function on two publicly available breast and colon cancer datasets having varying levels of imbalanced classes. The proposed CFAL function shows better discrimination abilities as compared to the popular loss functions such as ArcFace, CosFace, and Focal loss. Moreover, it outperforms several SOTA methods for histology image classification across both datasets.

R2-trans: Fine-grained visual categorization with redundancy reduction

Fine-grained visual categorization (FGVC) aims to discriminate similar subcategories, whose main challenge is the large intraclass diversities and subtle inter-class differences. Existing FGVC methods usually select discriminant regions found by a trained model, which is prone to neglect other potential discriminant information. On the other hand, the massive interactions between the sequence of image patches in ViT make the resulting class token contain lots of redundant information, which may also impact FGVC performance. In this paper, we present a novel approach for FGVC, which can simultaneously make use of partial yet sufficient discriminative information in environmental cues and also compress the redundant information in class-token with respect to the target. Specifically, our model calculates the ratio of high-weight regions in a batch, adaptively adjusts the masking threshold, and achieves moderate extraction of background information in the input space. Moreover, we also use the Information Bottleneck (IB) approach to guide our network to learn a minimum sufficient representations in the feature space. Experimental results on three widely-used benchmark datasets verify that our approach can achieve better performance than other state-of-the-art approaches and baseline models. The code of our model is available at: https://github.com/SYe-hub/R-2-Trans.

LVONet: automatic classification model for large vessel occlusion based on the difference information between left and right hemispheres.

Objective.Stroke is a highly lethal condition, with intracranial vessel occlusion being one of its primary causes. Intracranial vessel occlusion can typically be categorized into four types, each requiring different intervention measures. Therefore, the automatic and accurate classification of intracranial vessel occlusions holds significant clinical importance for assessing vessel occlusion conditions. However, due to the visual similarities in shape and size among different vessels and variations in the degree of vessel occlusion, the automated classification of intracranial vessel occlusions remains a challenging task. Our study proposes an automatic classification model for large vessel occlusion (LVO) based on the difference information between the left and right hemispheres.Approach.Our approach is as follows. We first introduce a dual-branch attention module to learn long-range dependencies through spatial and channel attention, guiding the model to focus on vessel-specific features. Subsequently, based on the symmetry of vessel distribution, we design a differential information classification module to dynamically learn and fuse the differential information of vessel features between the two hemispheres, enhancing the sensitivity of the classification model to occluded vessels. To optimize the feature differential information among similar vessels, we further propose a novel cooperative learning loss function to minimize changes within classes and similarities between classes.Main results.We evaluate our proposed model on an intracranial LVO data set. Compared to state-of-the-art deep learning models, our model performs optimally, achieving a classification sensitivity of 93.73%, precision of 83.33%, accuracy of 89.91% and Macro-F1 score of 87.13%.Significance.This method can adaptively focus on occluded vessel regions and effectively train in scenarios with high inter-class similarity and intra-class variability, thereby improving the performance of LVO classification.

Real-Time Pavement Crack Detection Based on Artificial Intelligence

Pavement as a structural element is closely related to the road traffic system because it is essentially a support structure for the movement of vehicles. Modern technology has attained significant improvements in road durability, quality, and safety through the use of new materials and construction technology. All civil engineering infrastructures have a specific lifespan. The road surface can be the worst, and repair is more costly if the pavement is not maintained. The objective of this research is to spot the damages over the inspected road and detect cracks in a short time, useful to achieve more comprehensive monitoring and assessment of the condition of road pavement using a smartphone equipped with You Only Look Once (YOLO) to reduce maintenance costs. The method uses advanced image processing techniques and uses YOLO to detect cracked pavement in a short time. More specifically, it is based on the latest generation of Deep Neural Network (DNN) algorithms, such as YOLO V5. YOLO V5 is used as a detector for cracked pavement and image processing as an automated pavement distress detector. The obtained result for longitudinal cracks shows that the area of cracks is 1.1682 m2, with a threshold value of 0.91, while the result for transversal cracks shows the area of cracks of 1.9627 m2, with a threshold value of 0.85. Otsu’s method works best under conditions such as low noise level, homogeneous lighting, and higher intra-class variance than an inter-class variance. The outcome of this project is improved recognition accuracy compared to the manual recognition currently used.

MSE-Net: A novel master–slave encoding network for remote sensing scene classification

Remote sensing scene (RSS) image classification plays a vital role in various fields such as urban planning and environmental protection. However, due to higher inter-class similarity and intra-class variability, achieving accurate classification for RSS images poses a considerable challenge for current convolutional neural networks (CNNs)-based and visual transformer (ViT)-based methods. To address these issues, this paper proposes a novel dual-encoding method named master–slave encoding network (MSE-Net) from two perspectives of feature extraction and fusion. The master encoder, based on ViT, extracts higher-level semantic features, while the slave encoder, based on CNN, captures relative lower-level spatial structure information. Secondly, to integrate feature information from the two encoders effectively, this paper further develop two fusion strategies. The first strategy involves the auxiliary enhancement units (AEUs), which eliminates semantic divergence between the two encoders, enhances spatial context awareness of the slave encoder and promotes effective feature learning. The interactive perception unit (IPU), as the second strategy, facilitates interaction and integration of the two encoders’ representations to extract more discriminative feature information. In addition, we conducted comparative experiments on four widely-used RSS datasets, including RSSCN7, SIRI-WHU, the aerial image dataset (AID) and NWPU-RESISC45 (NWPU45), to verify the effectiveness of MSE-Net. The experimental results demonstrate that MSE-Net achieved state-of-the-art (SOTA) performance across all the datasets.

Automated Hyperspectral Feature Selection and Classification of Wildlife Using Uncrewed Aerial Vehicles

Timely and accurate detection and estimation of animal abundance is an important part of wildlife management. This is particularly true for invasive species where cost-effective tools are needed to enable landscape-scale surveillance and management responses, especially when targeting low-density populations residing in dense vegetation and under canopies. This research focused on investigating the feasibility and practicality of using uncrewed aerial systems (UAS) and hyperspectral imagery (HSI) to classify animals in the wild on a spectral—rather than spatial—basis, in the hopes of developing methods to accurately classify animal targets even when their form may be significantly obscured. We collected HSI of four species of large mammals reported as invasive species on islands: cow (Bos taurus), horse (Equus caballus), deer (Odocoileus virginianus), and goat (Capra hircus) from a small UAS. Our objectives of this study were to (a) create a hyperspectral library of the four mammal species, (b) study the efficacy of HSI for animal classification by only using the spectral information via statistical separation, (c) study the efficacy of sequential and deep learning neural networks to classify the HSI pixels, (d) simulate five-band multispectral data from HSI and study its effectiveness for automated supervised classification, and (e) assess the ability of using HSI for invasive wildlife detection. Image classification models using sequential neural networks and one-dimensional convolutional neural networks were developed and tested. The results showed that the information from HSI derived using dimensionality reduction techniques were sufficient to classify the four species with class F1 scores all above 0.85. The performances of some classifiers were capable of reaching an overall accuracy over 98%and class F1 scores above 0.75, thus using only spectra to classify animals to species from existing sensors is feasible. This study discovered various challenges associated with the use of HSI for animal detection, particularly intra-class and seasonal variations in spectral reflectance and the practicalities of collecting and analyzing HSI data over large meaningful areas within an operational context. To make the use of spectral data a practical tool for wildlife and invasive animal management, further research into spectral profiles under a variety of real-world conditions, optimization of sensor spectra selection, and the development of on-board real-time analytics are needed.

Category attention guided network for semantic segmentation of Fine-Resolution remote sensing images

The semantic segmentation task is an essential issue in various fields, including land cover classification and cultural heritage investigation. The CNN and Transformer have been widely utilized in semantic segmentation tasks due to notable advancements in deep learning technologies. However, these methodologies may not fully account for remote sensing images' distinctive attributes, including the large intra-class variation and the small inter-class variation. Driven by it, we propose a category attention guided network (CAGNet). Initially, a local feature extraction module is devised to cater to striped objects and features at different scales. Then, we propose a novel concept of category attention for remote sensing images as a feature representation of category differences between pixels. Meanwhile, we designed the Transformer-based and CNN-based category attention guided modules to integrate the proposed category attention into the global scoring functions and local category feature weights, respectively. The network is designed to give more attention to the category features by updating these weights during the training process. Finally, a feature fusion module is developed to integrate global, local, and category multi-scale features and contextual information. A series of extensive experiments along with ablation studies on the UAVid, Vaihingen, and Potsdam datasets indicate that our network outperforms existing methods, including those based on CNN and Transformer.