Search

To address the landslide recognition problem in remote sensing images, this paper designs a visual transformer network model based on DEM (digital elevation model) feature enhancement, which is experimentally validated on the Bijie landslide dataset and Landslide4Sense2022 dataset. The lion optimizer is used during training. The results show that 98.49% accuracy and 97.24% F1 score are achieved on Bijie dataset, and 88.22% accuracy and 90.16% F1 score on Landslide4Sense2022 dataset, which is a significant improvement in landslide recognition compared with other mainstream network models. Therefore, it can be found that this paper's method is effective in the recognition of landslide from remote sensing images. Firstly, the swin transformer network model was successfully applied to remote sensing landslide image classification by means of transfer learning. Secondly, an information enhancement approach based on DEM features was designed. [ABSTRACT FROM AUTHOR]

While artificial intelligence technologies are increasingly studied as drivers of "digital authoritarianism," resistance to this process has remained underexplored. Our paper addresses this gap by asking why and how citizens resist AI-powered autocratization in consolidated democracies. We first conceptualize this resistance as distinct from other forms of anti-autocratic resistance in that it is anticipatory rather than reactive, and directed at existing authorities rather than new democratic challengers. We then introduce techno-authoritarian imaginaries as a novel concept to understand the drivers and shapes of this resistance. First, we argue that activists from civil society draw on these broader, pre-existing imaginaries of authoritarian futures to make sense of new technologies and articulate technology-specific problem frames. Second, we propose that imaginaries are contingent on historical and political experiences and therefore differ between contexts. Such differences, in turn, shape how the respective targets respond to resistance. We illustrate our argument by case studies of campaigns against facial recognition technology in the U.S. and the European Union. Our paper enriches existing debates on resistance to autocratization and advocates or a more pronounced engagement with practices of future-making, as constructions of societal futures—both desirable and undesirable—are becoming an increasingly important source for democratic mobilization. [ABSTRACT FROM AUTHOR]

Environmental applications play a significant role in the ongoing research area of Polarimetric Synthetic Aperture Radar (PolSAR) image classification. In this paper, a new model is proposed for classifying PolSAR images and applied to a part of the Greater Cairo area in the Nile basin, South of Delta, Egypt. First, the proposed model performs data pre-processing by extracting the coherency and covariance elements noted as [T] and [C] matrices, respectively. Second, temporal convolutional networks (TCN) deep learning is used to extract the features from coherency and covariance elements and then train the model. Third, the SoftMax classifier is used to classify the PolSAR image. Finally, the proposed model is tested with evaluation metrics. The obtained results show that the proposed model can achieve high classification performances. [ABSTRACT FROM AUTHOR]

A pulse-like ground motion identification method is proposed based on the Siamese Convolutional Neural Networks (SCNNs). The wavelet coefficient graphs of pulse-like ground motions are used as the input data, and the features extracted by the SCNNs are used for the identification. Based on the time-domain features, pulse-like and non-pulse-like ground motions are classified into several classes. The results indicated that all the identified pulse-like ground motions have similar pulse features to the pre-selected training data. The principal advantage of the method is that the misclassification problem can be minimized by an updating procedure proposed by this paper. [ABSTRACT FROM AUTHOR]

Classification of hyperspectral image (HSI) based on deep learning has attracted extensive attention. However, there is a strong spatial nonlinear coupling phenomenon among data in HSI, which leads to low spatial separability of different types of data and affects classification accuracy. In order to solve this problem, this paper proposes a HSI classification method based on nonlinear spatial decoupling strategy and deformable convnets v2 (DCNv2). Specifically, this paper introduces the classic color names (CN) algorithm with strong nonlinear projection ability in the field of target tracking. Through the projection matrix of CN algorithm, the spatial constraint features of the original spectral data extracted by bilinear local binary pattern and the original spectral data are projected to realize the spatial decoupling of features and improve the spatial separability of data. Then, the DCNv2, which has strong adaptability of geometric transformation, is used to extract the deep features of the decoupled data, so as to achieve effective classification. The experiment was carried out on four standard hyperspectral data sets and the overall accuracy of the proposed method on these data sets reached 99.73%, 99.93%, 99.98% and 99.99%, respectively. Compared with several classical classification methods, it has better classification effect. [ABSTRACT FROM AUTHOR]

The classification of multi- and hyper-spectral imagery is essential for several remote sensing applications. However, several challenges still hinder the accuracy and efficiency of learning algorithms. In diverse remote sensing applications, the high spectral heterogeneity between and within classes is often an issue that adds uncertainties to the classification. Most classifiers do not consider the influence of non-target classes and variations in the spectral similarity level between classes during classification. This paper presents a novel learning strategy for remote sensing image classification, which exploits non-target classes to enhance the generation of classification masks and the performance of supervised methods for specific classes of interest. It produces multiple classification instances, based on the incorporation of counterexample information into the training data set, in order to refine the separation of classes of interest in the available feature space. The experimental results demonstrate the potential of the proposed counterexample assisted learning for road extraction, using high spatial-resolution imagery data. [ABSTRACT FROM AUTHOR]

Modern manufacturing is the world's largest and most automated industrial sector. The rise of Industry 4.0 technologies such as Big Data, Internet of Things (IoT) devices, and Machine Learning has enabled a better connection with machines and factory systems. Data harvesting allowed for a more seamless and comprehensive implementation of the knowledge-based decision-making process. New models that provide a competitive edge must be created by combining the Lean paradigm with the new technologies of Industry 4.0. This paper presents novel computer-based vision models for automated detection and classification of damaged packages from intact packages. In high-volume production environments, the package manual inspection process through the human eye consumes inordinate amounts of time poring over physical packages. Our proposed three different computer-based vision approaches detect damaged packages to prevent them from moving to shipping operations that would otherwise incur waste in the form of wasted operating hours, wasted resources and lost customer satisfaction. The proposed approaches were carried out on a data set consisting of package images and achieved high precision, accuracy, and recall values during the training and validation stage, with the resultant trained YOLO v7 model. [ABSTRACT FROM AUTHOR]

As the core technology of precision animal husbandry, efficient and rapid identification of Jinnan cattle individuals can promote the scale, informatization and refinement of breeding, which is very necessary for the development of animal husbandry at this stage. However, the traditional livestock individual recognition method based on ear tag is labour-consuming, time-consuming, inefficient, easy to wear and limited by the recognition distance, and the accuracy is also very low. In order to solve this problem, a new method of Jinnan cattle individual recognition based on deep mutual learning is proposed by using the non-contact image recognition method. Two student networks are designed. They supervise each other and complete the task together. Their efficiency can be higher than that of a strong teacher network. Through this method, the individual recognition performance of Jinnan cattle is also enhanced. The experimental results verify the effectiveness of the method of deep mutual learning. Consult and learn from the peer network are used to improve generalization, so as to improve model recognition performance. Finally, the accuracy is 99.3% on the Jinnan cattle individual dataset established in this paper. The application of the contactless cattle individual recognition method in the farm is of great significance. [ABSTRACT FROM AUTHOR]

Histopathological images of the breast play a crucial role in diagnosing breast cancer and developing effective treatment strategies. However, the complexity of these images poses challenges for traditional convolutional neural networks, which struggle to extract all relevant features. In this paper, we present a novel classification method that addresses this limitation and enhances the accuracy of diagnosing benign and malignant breast tumors. Our approach involves data augmentation techniques, such as staining normalization, image patch generation, and spatial geometry transformation, to expand the training set. We utilize BCMNet (Breast Classification Fusion Multi-Scale Feature Network), which combines VGG16 and CBAM, to fuse spatial features, channel features, and multi-scale features of the input images. We evaluate our method on the BreakHis dataset and achieve average accuracies of 91.91%, 91.14%, 92.65%, and 87.56% for breast histopathology image classification at four different magnifications. Compared to the benchmark network VGG16 without data augmentation, our approach improves the highest accuracy by 3.69% and the average accuracy by 18.47%. Moreover, our proposed method achieves an F1 score of 95.02% and an average recognition accuracy of 92.20% at the patient level. Our approach surpasses previous breast pathology image classification methods, both single and hybrid networks, and establishes a strong theoretical foundation for computer-assisted breast cancer diagnosis. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Natural Fibers is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Identity recognition technology is a type of technology that realizes identity verification based on certain biological characteristics. After entering the Internet era, this technology has become a popular research direction in the computer field. In this paper, the image of the tooth print is used as the biological feature to carry out the research on the identification algorithm. This paper adopts the target detection algorithm based on neural network to detect a single tooth imprint area of the target, build a target detection network. The experimental results show that the method has a good segmentation effect on the target area, and the accuracy rate is 91.66%. According to the contour features of the collected tooth print images, a set of tooth pore area ratio feature extraction methods are designed. To objectively evaluate the recognition and classification method, the support vector machine is used as the final classifier. The recognition accuracy rate is 94.09%, and the verification accuracy rate is 94.09%. The test accuracy rate is 91.46%, and the classification effect is excellent. This paper has made a lot of breakthroughs and obvious progress based on the previous research on the tooth impression model. [ABSTRACT FROM AUTHOR]

In this paper, we present automatic, deep-learning methods for pipeline detection in underwater environments. Seafloor pipelines are critical infrastructure for oil and gas transport. The inspection of those pipelines is required to verify their integrity and determine the need for maintenance. Underwater conditions present a harsh environment that is challenging for image recognition due to light refraction and absorption, poor visibility, scattering, and attenuation, often causing poor image quality. Modern machine-learning object detectors utilize Convolutional Neural Network (CNN), requiring a training dataset of sufficient quality. In the paper, six different deep-learning CNN detectors for underwater object detection were trained and tested: five are based on the You Only Look Once (YOLO) architectures (YOLOv4, YOLOv4-Tiny, CSP-YOLOv4, YOLOv4@Resnet, YOLOv4@Den seNet), and one on the Faster Region-based CNN (RCNN) architecture. The models' performances were evaluated in terms of detection accuracy, mean average precision (mAP), and processing speed measured with the Frames Per Second (FPS) on a custom dataset containing underwater pipeline images. In the study, the YOLOv4 outperformed other models for underwater pipeline object detection resulting in an mAP of 94.21% with the ability to detect objects in real-time. Based on the literature review, this is one of the pioneering works in this field. [ABSTRACT FROM AUTHOR]

Early cost estimation is a decisive value driver in the product development process in manufacturing industries. Machine learning offers new intelligent methods to support traditional cost calculation processes. While traditional research on intelligent cost estimation focuses on machine learning regression or classification models, we propose a new approach based on interlocking deep learning methods. In this paper we investigate the applicability of deep learning techniques, focusing on image recognition and deep learning regression as well as autoencoding to estimate product costs of circuit boards to be purchased. We create and evaluate deep learning models using real-world data from an original equipment manufacturer (OEM). Our findings suggest that deep learning models can streamline cost calculation and estimation processes while deep learning object recognition-based cost estimation outperforms autoencoding techniques. This research is designed to be transferable to other cost estimation projects. [ABSTRACT FROM AUTHOR]

Remote sensing image (RSI) scene classification is a hot topic in the field of remote sensing and has garnered a lot of attention. The key issue in image classification is effectively understanding semantic content. Convolutional neural networks (CNNs) are generally recognized to significantly improve classification performance due to their powerful feature extraction capabilities. However, the overall structure of the model is complicated and has a large number of parameters, making it difficult to extract more efficient features. To address these problems, in this paper, we propose a lightweight skip-connected expansion Inception network called SEINet. To capture characteristics at a more granular level, we create a new lightweight backbone network with fewer parameters based on the existing network architecture. Additionally, the paper introduces a skip-connected expansion Inception (SEI) module for extracting context-dependent relationships. The ablation experiments verify the effectiveness of our proposed module. Experiment findings on two public datasets demonstrate that our method has advantages in classification accuracy and execution efficiency over state-of-the-art (SOTA) methods. [ABSTRACT FROM AUTHOR]

In recent years, methods of spatial-spectral feature extraction have been widely used in hyperspectral image (HSI) classification and have achieved good performance. However, the ways are often unable to distinguish effectively the boundary of the adjacent ground objects. Multi-scale feature fusion alleviates the problems to a certain extent, but in the case of limited samples, the previous methods often fail to achieve the desired effect. In this paper, we propose a multi-scale hyperspectral image classification method based on weakened Laplacian pyramid and guided filtering to nicely distinguish the boundary of the different ground objects. This structure first obtains multi-scale information by constructing a weakened Laplacian pyramid, and weakens the magnification of Gaussian blur in the upsampling process. While obtaining multi-scale images, it can not only retain enough image information to expand feature samples but also emphasize edge information in the edge extraction of the Laplacian pyramid. Then, the weakened Laplacian pyramid feature map is used as the guiding image of the guiding filter, which can better preserve the edge features and further improve the classification performance of HSI. Extensive experiments are carried out on four HSI data sets. The results show that our method is superior in classification accuracy and computational cost. [ABSTRACT FROM AUTHOR]

High Spatial and Spectral Resolution (HSSR) remote-sensing images can provide rich spectral bands and detailed ground information, but there is a relative lack of research on this new type of remote-sensing data. Although there are already some HSSR datasets for deep learning model training and testing, the data volume of these datasets is small, resulting in low classification accuracy and weak generalization ability of the trained models. In this paper, an HSSR dataset Luojia-HSSR is constructed based on aerial hyperspectral imagery of southern Shenyang City of Liaoning Province in China. To our knowledge, it is the largest HSSR dataset to date, with 6438 pairs of 256 × 256 sized samples (including 3480 pairs in the training set, 2209 pairs in the test set, and 749 pairs in the validation set), covering area of 161 km2 with spatial resolution 0.75 m, 249 Visible and Near-Infrared (VNIR) spectral bands, and corresponding to 23 classes of field-validated ground coverage. It is an ideal experimental data for spatial-spectral feature extraction. Furthermore, a new deep learning model 3D-HRNet for interpreting HSSR images is proposed. The conv-neck in HRNet is modified to better mine the spatial information of the images. Then, a 3D convolution module with attention mechanism is designed to capture the global-local fine spectral information simultaneously. Subsequently, the 3D convolution is inserted into the HRNet to optimize the performance. The experiments show that the 3D-HRNet model has good interpreting ability for the Luojia-HSSR dataset with the Frequency Weighted Intersection over Union（FWIoU） reaching 80.54%, indicating that the Luojia-HSSR dataset constructed in this paper and the proposed 3D-HRnet model have good applicable prospects for processing HSSR remote sensing images. [ABSTRACT FROM AUTHOR]

The automated process for dental caries detection draws increasing attention with the technological innovation in machine learning methods. This is a core issue in dental diseases especially in the detection of caries as it leads to serious health ailments. This paper takes an effort to adequately segment and identify dental diseases. There are four main steps. The preprocessing technique uses binary histogram equalization which increases the texture region visibility for the caries detected on dental images. The novel technique of segmentation with Curvilinear Semantic Deep Convolutional Neural Network (CSDCNN) is proposed in this paper. The segmentation is followed by the proposed Inception resnetV2, which acts as the classification technique to determine the caries in dental images. The proposed segmentation algorithm is used to determine a dental degree of membership. The inception is brought out with different scales of information, which relates to various input images as data. An examination of the x-ray images will detect the impact of illness on a tooth. Particularly for the segmentation and classification mission, we deemed four diseases: dental caries, periapical infection, periodontal, and pericoronal diseases. Based on the number of input functional parameters, the Inception resnetV2 classifies different image categories effectively. The proposed Inception resnetV2 has become the most effective tool in machine learning to solve problems like image classification with a high order of accuracy. The average accuracy of the device proposed is 94.51%. This provides higher classification accuracy when compared to other existing methods. [ABSTRACT FROM AUTHOR]

Seagrass beds play a vital ecological role by regulating climate, maintaining biodiversity, and sequestering carbon. Achieving accurate classification of seagrass beds across both time and space is of significant importance for efficient monitoring. This paper addresses the challenges of seagrass bed classification in remote sensing imagery and the difficulty of achieving temporal and spatial transferability of classification model. Leveraging a long-term Landsat satellite image time series spanning from 1976 to 2022 in the Yellow River Delta, we enhance the DeepLab V3+ network and propose a cross-temporal-spatial domain seagrass bed transfer learning classification approach. We utilize selectively chosen images from a few years in partial regions for training, introduce freeze-training and-unfreeze-training during the model training phase, and then transfer the model to achieve cross-temporal and cross-temporal-spatial domain classification on images from other years and regions. Additionally, we discuss the factors influencing the classification results. The research results indicate that: (1)The method we proposed has shown good results, achieving classification accuracy of 83.17% and 80.99% for seagrass beds in both cross-temporal and cross-temporal-spatial domains. (2) The classification results indicate that this method achieves an average classification accuracy higher by 16.18% compared to networks such as UNet, PSPNet, HRNet, UNet++, SegFormer, etc. under cross-temporal conditions and higher by 17.56% under cross-spatiotemporal conditions for seagrass bed classification. (3) Based on the classification results, it is observed that from 1973 to 1986, the overall area of seagrass beds in the Yellow River Delta had an initial growth trend, with the largest seagrass bed area recorded in 1986, reaching 4281.289 hectares. Subsequently, the seagrass beds showed an overall decreasing trend, and by 2019 they had almost completely disappeared. The cross-temporal-spatial domain seagrass bed transfer learning classification method proposed in this paper can provide valuable support for the monitoring, restoration, management, and utilization of seagrass ecosystems. [ABSTRACT FROM AUTHOR]

Classification of hyperspectral image (HSI) is an essential step in various remote sensing tasks. Currently, transformer-based models have made significant contributions to HSI classification due to their remarkable ability to model the relation among long-range sequence data. However, HSI data exhibit strong nonlinear coupling and strong correlation, leading to a decrease in the classification performance of transformer-based models. The SpectralFormer network has achieved excellent performance in HSI classification, but its cross-layer adaptive fusion mechanism cannot effectively highlight critical information, affecting the representation ability of spectral-spatial features. To address the above issues, this paper proposes an HSI classification method based on local feature decoupling and SpectralFormer with a hybrid attention module (HAMSF) network. Specifically, this paper first uses the histogram of oriented gradient algorithm to preprocess HSI coupled data and achieve preliminary nonlinear decoupling of HSI data. Secondly, the hybrid attention module is embedded in a cross-layer adaptive fusion mechanism, generating refined neighbouring information before each information fusion to improve the spectral-spatial feature representation ability. Finally, the processed data are input into the HAMSF network for the final decoupling and classification of HSI. The experiment is carried out on three benchmark hyperspectral datasets, and the experimental results confirm that the proposed method has better classification performance compared to several transformer-based and classical HSI classification methods. [ABSTRACT FROM AUTHOR]

HyperGraph Neural Network (HGNN) has recently emerged as a promising approach for hyperspectral image classification (HSIC), reconciling state-of-the-art performance with powerful representation capabilities. However, existing HGNN-based methods have limited ability for hypergraph structure exploitation, leading to imperfect classification results. In this paper, we propose a framework called the residual enhanced hypergraph Neural Network (ResHGNN) to discover the potential structural features in hyperspectral image (HSI) data during deep neural networks. Specifically, ResHGNN first generates hyperedges from spatial-spectral features to construct a hypergraph representing fused spatial-spectral feature relationships in HSI. Then, the higher-order relationship among fused modal features is optimized by a residual enhanced hypergraph convolution learning process, to circumvent the HGNN-related over-smoothing issue. Experiments over three popular hyperspectral datasets show that the proposed classification method yields better performance than other models on the visual and numerical comparison. [ABSTRACT FROM AUTHOR]

Recently, convolutional neural network (CNN) and Transformer have been successfully applied in hyperspectral image (HSI) classification. Despite their success, some challenges persist. Popular HSI classification methods typically employ only a single patch as the network input, which may not provide robust information for various datasets or within different scenarios of a specific dataset. Large-scale patches are usually conducive to classifying large-scale scenes, while small-scale patches excel at classifying boundary locations. In addition, the majority of methods do not distinguish between the central pixel and its adjacent pixels, although the information from the central position is crucial. In this paper, a novel double-branch hierarchical partial convolution network (DBHP-Net) is proposed to address the aforementioned challenges. Concretely, DBHP-Net employs a new convolution, named TConv, to emphasize mining the spatial-channel features of the central position. Then, the TConvFormer Block is built by adopting the advanced engineering techniques of Transformer. Furthermore, we construct our model in a hierarchical style. The double-branch structure with large-scale and small-scale patches provides complementary information from the dataset. Experimental results based on randomly selected samples and spatially disjoint samples of two benchmark datasets demonstrate that our method outperforms other state-of-the-art (SOTA) methods. [ABSTRACT FROM AUTHOR]

In recent years, some hyperspectral image (HSI) classification methods based on deep models have shown excellent performance. Most deep models receive three-dimensional (3D) block structures as input to extract spectral-spatial features from HSI data. However, the existence of background pixel information in 3D blocks affects the spectral-spatial feature representation ability. In addition, directly performing overall dimension reduction on high-dimensional spectral information lacks analysis of the original data, which weakens the constraint relationship between spectral data and increases the difficulty of feature extraction. Therefore, an HSI classification method based on feature enhancement and hybrid deformable convolution network is proposed. Firstly, a novel 3D block is constructed to enhance spectral-spatial feature representation ability by replacing background pixel information with non-background pixel information. Secondly, based on the analysis results of the original spectral data, this paper proposes a regional dimension reduction mechanism to enhance the constraint relationship between HSI data. Finally, the proposed method is applied to hybrid deformable convolution network for experimental verification. The experimental results of three standard datasets show that the classification effect of the proposed method is better than that of several classification methods, which proves the effectiveness of the proposed method. [ABSTRACT FROM AUTHOR]

Classification with a few samples of training set has been a long-standing issue in the field of polarimetric synthetic aperture radar (PolSAR) image analysis and processing. In fact, one of the most important challenges of the PolSAR image classification task is the number of labelled samples. In essence, training network by utilizing just few numbers of training samples cannot be led to the reliable result since the neural network is so sensitive to the number of training samples. In addition, in the real PolSAR image classification task, the huge number of training samples is not accessible. On the other hand, classification performance using just small number of labelled samples is not an accurate result. So, in this paper, aiming at the small number of training samples of the PolSAR image classification task, a novel ensemble self-supervised feature-learning (ESSFL) model is designed. The designed ESSFL can automatically extract PolSAR features conducive to PolSAR image classification with a small number of training samples. In addition, it can significantly decrease the dependence of neural network algorithms on large labelled samples of training set. First, to utilize the spatial–polarimetric features of PolSAR data perfectly, the EfficientNet-B0 is presented and utilized as the main section of the deep learning (DL) model to extract DL features of PolSAR data. Then, using an optimization function that constrains the cross-correlation matrix of various distortions of each sample to the identity matrix, the designed deep learning model can obtain the effective features of homogeneous samples gathering together and heterogeneous samples separating from each other in a self-supervised manner. Moreover, two ensemble learning (EL) models, feature-level and view-level ensemble, are proposed to increase the feature extraction capability and classification result by jointly using spatial features at different scales and polarimetric information at different bands. Finally, the stack of the obtained features and the main polarimetric information of PolSAR data can be input into classifier for the classification of PolSAR data. In this paper we use two different classifiers, the first one is the Random Forest (RF) classifier and the second one is the Support Vector Machine (SVM) classifier. The advantages of SVM include that it can be used to avoid the difficulties of using linear functions in the high-dimensional feature space, and the optimization problem is transformed into dual convex quadratic programs that are so usable for an actual image classification task. In addition, among all the available classification methods, random forests provide one of the highest accuracies. The random forest technique can also handle big data with numerous variables running into thousands. In fact, it can automatically balance data sets when a class is more infrequent than other classes in the data that is so important for a real PolSAR image classification task. Experimental results on three well-known PolSAR data sets illustrate that the designed ESSFL can extract more discriminant features using the designed deep learning model. In the end, the experiments prove that the designed ESSFL has a significant classification performance compared with different deep learning models in the case of small number of training samples, and also it can achieve a better result in the case of large number of training samples by comparison with the most of the deep learning models. [ABSTRACT FROM AUTHOR]

Copyright of Canadian Journal of Remote Sensing is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Multilabel remote sensing image classification (MLRSIC) has received increasing research interest. Taking the co-occurrence relationship of multiple labels as additional information helps to improve the overall performance. However, current methods only focus on using it to constrain the final feature which is output from a convolutional neural network (CNN). On the one hand, these methods need to exploit the potential of label correlation in feature representation fully. On the other hand, they increase the label noise sensitivity of the system, resulting in poor robustness. In this paper, a novel method called 'Semantic Interleaving Global chaNnel Attention' (SIGNA) is proposed for MLRSIC. First, the label co-occurrence graph is obtained according to the statistical information of the training set and fed into a graph neural network (GNN) to generate optimal semantic feature representations of each label. Next, the semantic features are interleaved with visual features which are extracted by CNNs to guide the overall features of the input image transform from the original feature space to the semantic feature space with embedded label relations. Then, global attention triggered by semantic interleaving is used to emphasize visual features in important channels. Finally, to make SIGNA easier to use and more optimized, multihead SIGNA-based feature adaptive weighting networks are proposed as plug-in blocks to plug into any layers of a CNN. For remote sensing images, better classification performance can be achieved by inserting the plug-in blocks into the shallow layers of CNNs. We conducted extensive experimental comparisons on three data sets: UCM, AID and DFC15. Experimental results demonstrate that the proposed SIGNA achieves superior classification performance compared to state-of-the-art (SOTA) methods. Notes that the codes of this paper will be open to the community for reproducibility research. [ABSTRACT FROM AUTHOR]

For the particular engineering-geological conditions and natural environment in permafrost regions, the highway has a high damage ratio during the operation. In warm and ice-rich permafrost regions, highway pavement and embankment have poor serviceability, which threatens driving the vehicle and safe operation. Traditional field surveys on highway damage cannot entirely, rapidly and precisely obtain the distress information along the whole road. In this paper, the Qinghai-Tibet Highway (QTH) pavement distress was investigated in 550-km-long permafrost regions based on the field surveys in 2014 and 2019 and the unmanned aerial vehicle (UAV) imagery. The type and damage ratio of distress were extracted by the remote sensing image classification method. The pavement roughness and embankment height were acquired by the spatial analysis of geographical information system. This paper provides a novel insight and efficient method for the distress investigation and exploration of distress formation in permafrost region, which can be applied in other similar engineering projects in the cold and inclement permafrost region. Furthermore, this paper also presents valuable and first-hand field data for evaluating highway serviceability and prevention of road damage during operation and maintenance stages. [ABSTRACT FROM AUTHOR]

Remote sensing image scene classification is a fundamental task in intelligent interpretation of remote sensing images. Although Transformers possess a powerful attention mechanism, they require lengthy training procedures to achieve good performance levels. To address this issue, this paper proposes a novel deep learning network model by combining CNN and Swin Transformer named SwinHCST. Firstly, the model uses Weighted Normalized CNN to quickly extract low-level features of the image. Secondly, the Receptive Field Block module facilitates multi-scale information fusion, Thirdly, the Information Fusion Transformer further excavates the deep-level features of the image. Furthermore, this paper has designed a plug-and-play Cross Spatial Information Fusion Block, which is used to encodes dimensional information and extracts global information to enhance information exchange. The scene classification experiments show that the proposed model outperforms other methods on the three selected datasets and can achieve excellent performance without requiring large amounts of data and training. Specifically, the classification accuracy of the proposed method on the three datasets is 93.76%, 93.60%, and 98.10%, which is 1.7% to 3.71% higher than ResNet50 and 3.7% to 5.7% higher than Swin Transformer. [ABSTRACT FROM AUTHOR]

Land-cover type interpretation by the use of remote sensing image classification techniques is always a hot topic. In this paper, an object-oriented method is presented for fully polarimetric synthetic aperture radar (SAR) image classification. Differing from most of the traditional object-oriented classification algorithms, the proposed method employs an innovative classification strategy that combines a pixel-based classifier and a region growing technique. Firstly, taking each individual pixel as a seed pixel, the homogeneous areas are extracted by a region growing technique. Then, using the information of the pixel-based classification result, the pixels located in each homogeneous area are all assigned to a certain class. Finally, the majority voting strategy is deployed to determine the final class label of each pixel. The experiments conducted on two fully polarimetric SAR images reveal that the proposed classification scheme can obtain pleasing classification accuracy and can provide the classification maps with more homogeneous regions than pixel-based classification. [ABSTRACT FROM AUTHOR]

In this paper, a multi-strategy fusion (MSF) framework, based on improved MBLBP and bi-exponential edge-preserving smoother (BEEPS), is proposed for hyperspectral image (HSI) classification. First, MBLBP operator is adopted to characterize the overall structural information of HSI, where the averaging strategy allocates same weights for the pixels in a local sub-region, so that the edges tend to be blurred due to being isotropic. To solve this question, the steering kernel is first introduced into MBLBP for learning the local structure prior of HSI. Then, a support vector machine classifier is used to calculate the soft classified probabilities of pixels. Furthermore, BEEPS is adopted to smooth the soft classified probabilities maps in the post-processing stage, and the purpose is to further improve classification accuracy of HSI by considering context-aware information for each class label. Experiments are performed on three real hyperspectral datasets, namely, Indian Pines, KSC, and Houston 2013, only 1%, 6, and 5 labeled samples are randomly selected for training, the overall accuracy(kappa) obtained by MSF is 99.47%(99.40), 99.52%(99.47), and 94.25%(93.78), respectively, which is far better than the contrast methods. [ABSTRACT FROM AUTHOR]

In recent years, deep learning methods have been widely used for the classification of hyperspectral images. However, their limited availability under the condition of small samples remains a serious issue. Moreover, the current mainstream approaches based on convolutional neural networks do well in local feature extraction but are also restricted by its limited receptive field. Hence, these models are unable to capture long-distance dependencies both on spatial and spectral dimension. To address above issues, this paper proposes a global-local U-Net augmented by transformers (TrmGLU-Net). First, whole hyperspectral images are input to the model for end-to-end training to capture the contextual information. Then, a transformer-augmented U-Net is designed with alternating transformers and convolutional layers to perceive both global and local information. Finally, a superpixel-based label expansion method is proposed to expand the labels and improve the performance under the condition of small samples. Extensive experiments on four hyperspectral scenes demonstrate that TrmGLU-Net has better performance than other advanced patch-level and image-level methods with limited training samples. The relevant code will be opened at [ABSTRACT FROM AUTHOR]

This paper introduces a lightweight Convolutional Neural Networks (CNN) method for image classification in COVID-19 diagnosis. The proposed approach emphasizes simplicity while achieving high performance, and it leverages a meticulously annotated database. The CNN model consists of four convolutional layers, followed by flattening and two dense layers. The methodology focuses on classifying 2D slices of Computed Tomography (CT) scans. To enhance accuracy, the slices undergo anatomy-relevant masking and the removal of non-representative slices from the CT volume. This is achieved by cropping a fixed-sized rectangular area to capture the relevant region of interest and using a threshold based on bright pixels in binarized slices. The proposed methodology demonstrates improved quantitative results in slice classification by employing slice processing techniques. Additionally, augmentation techniques such as class weight balancing, slice flipping, and a learning rate scheduler are applied to diagnose at the slice level. For patient-level diagnosis, a majority voting method is employed by considering the slices of each CT scan. The proposed method surpasses the baseline approach and other alternatives in terms of macro F1 score, both on the validation set and a test partition containing previously unseen images from the rigorously annotated dataset. [ABSTRACT FROM AUTHOR]

In order to improve the harvesting efficiency of lodging crops, starting from the reel system, this paper combines intelligent image processing algorithms to identify the characteristics of lodging crops. Moreover, this paper optimises the data set, chooses L2 norm regularisation to reduce overfitting, uses exponential decay method to set learning rate, uses moving average model to speed up training convergence speed, improves classification accuracy, and simplifies the structure of the neural network to improve the real-time image recognition effect. A moving average is a statistical estimation that is used to evaluate sets of data by comparing the amounts of various subsets of the entire data set. Simultaneously, this paper improves the traditional reel structure and adds image recognition equipment to recognise crop lodging in real time during harvesting. In addition, this paper reasonably controls the reel system for harvesting, sets the reel working status by identifying lodging crops, and combines image technology to achieve efficient operation of the reel system. Finally, this paper designs experiments to verify the system performance. The research results show that the performance of the system constructed in this paper is good. [ABSTRACT FROM AUTHOR]

High-resolution remote sensing images provide complete and detailed ground scenes. However, these complex and diverse details make it difficult to interpret large area objects and their small area details at a single granularity. Therefore, the method proposed in this paper refers to the principles of visual working memory, visual attention and visual reasoning, independently analyse the ground objects of different granularities, and integrates the recognition results of the different granularity ground objects to form accurate classification results. This multi-granularity classification strategy suppresses noise interference, protects typical edges, and highlights clear details. Experimental results with dataset of real land use survey projects, including images of different time, multiple perspectives, and different altitudes, The results show that the idea of combining unsupervised edge detection with neural network classification can greatly enhance the generalization ability and edge cognitive ability, making the classification accuracy reach 80% on average, with an improvement of about 5%. [ABSTRACT FROM AUTHOR]

Hyperspectral images (HSIs) contain notable information of land objects by acquiring an immense set of narrow and contiguous spectral bands. Feature extraction (FE) and feature selection (FS) as dimensionality (band) reduction strategies are performed to enhance the classification result of HSI. Principal component analysis (PCA) is frequently exploited for the FE of HSI. However, it often possesses the inability to extract local and subtle HSI structures. As such, segmented-PCA (SPCA), spectrally segmented-PCA (SSPCA) and folded-PCA (FPCA) are presented for local and useful FE from the HSI. In this paper, we propose two FE methods called segmented-FPCA (SFPCA) and spectrally segmented-FPCA (SSFPCA). SFPCA exploits SPCA and FPCA while SSFPCA exploits SSPCA and FPCA together. In particular, SFPCA and SSFPCA apply FPCA on highly correlated and spectrally grouped HSI bands, respectively. We consider nonlinear methods Kernel-PCA (KPCA) and Kernel entropy component analysis (KECA) for extended comparison. For the experimented agricultural Indian Pine and urban Washington DC Mall HSIs, the results manifest that SFPCA (95.6262% for the agricultural HSI and 97.4782% for the urban HSI) and SSFPCA (96.3221% for the agricultural HSI and 98.0116% for the urban HSI) outperform the conventional methods. [ABSTRACT FROM AUTHOR]

In response to the call for implementing national waste classification, this paper proposes an intelligent waste classification system based on the improved MobileNetV3-Large, which can raise the national awareness of waste classification through the combination of software and hardware. The software module is based on WeChat applet and offers functions for image recognition, text recognition, speech recognition, points-based quiz and so on. The hardware module is based on Raspberry Pi and covers image shooting, image recognition, automatic classification with automatic announcement and so on. The algorithm model applied to the image classification adopts a network model based on MobileNetV3-Large. This network model is enabled to classify garbage images through deep separable convolution, inverse residual structure, lightweight attention structure and the hard_ swish activation function. The text classification model adopts a network model based on LSTM, extracts text features through word embedding, enhancing the effect of garbage text classification. After testing, the system can leverage deep learning to realise intelligent garbage classification. The image recognition accuracy of the algorithm model was found to reach 81%, while the text recognition accuracy was as high as 97.61%. [ABSTRACT FROM AUTHOR]

This paper presents a quantitative index introduced, ‘clarity,’ which can address both the ultrasound probe's brightness and full contact states to indicate ultrasound image quality for the robotic ultrasound system. Furthermore, it evaluates its performance through an experiment, a comparative analysis, and a questionnaire. The index is calculated based on pixel statistics within the predefined region of interest. Experimental validation is conducted by implementing a robotic ultrasound system and performing experiments using two phantoms. Notably, the Results demonstrate the distinction of the visible difference in ultrasound images but also show that it is hard to classify the tiny difference. Additionally, a questionnaire was carried out to ask about the recognition of good image quality. The opinions are the same trends and patterns. The 53.4 clarity value was determined from the answers as the boundary between good and bad-quality images. The experiment results, and questionnaire findings suggest that the proposed clarity index could be a reliable tool for evaluating ultrasound image quality in robotic systems, offering insights for improving diagnostic accuracy and efficiency. This research provides a clear framework for assessing ultrasound image quality in robotic applications, paving the way for advancements in the efficacy and precision of ultrasound imaging techniques. [ABSTRACT FROM AUTHOR]

The excellent capabilities of Transformers and Graph Neural Networks (GNNs) in modelling long-range dependencies and handling irregular data have led to their widespread application in hyperspectral image (HSI) classification tasks. However, the Graph Transformer combining both advantages is rarely used in this field and has some limitations. Current Graph Transformers consider interactions between all nodes within the graph, adding complexity and introducing unnecessary information from noisy nodes. Moreover, the rich spectral information in HSIs is often ignored, and there is a lack of effective fusion of spatial information. In this paper, we propose a dual-stream graph-guided Transformer for HSI classification. In spatial dimension, superpixels are utilized to guide spatial graph generation, capturing global topological dependencies and local details effectively in HSIs. In terms of spectrum, we innovatively construct a spectral graph based on spectral channels and adopt a contribution score-based strategy to adaptively filter out irrelevant edges, achieving sparsity while preserving spectral context relationships. Experimental results demonstrate the significant competitive advantage of our method in HSI classification tasks on three public datasets. The code is available at https://github.com/youngboy03/GTDPNet. [ABSTRACT FROM AUTHOR]

Considering that traditional image recognition methods for coal and gangue are susceptible to environmental factors such as light and dust, this paper proposes an image recognition and detection method for the coal and gangue based on infrared thermal imaging technology and an improved YOLOv8n. First, we established a thermal infrared imaging platform for coal and gangue. We used the YOLOv8n model to analyze the differences between coal and gangue at various temperatures, determining the optimal temperature for effective recognition of the coal and gangue. Next, to facilitate the practical deployment of the detection model for coal and gangue, we incorporated the Ghost module into the original YOLOv8n model to reduce its complexity and storage requirements. Unfortunately, incorporating the Ghost module led to a slight decrease in the model’s detection performance. Next, we introduced the DAttention module into the model to enhance its feature extraction capabilities. At the same time, we replaced the CIoU loss function with DIoU, which further improved the model’s performance metrics. Finally, we compared the improved YOLOv8n model with Faster-RCNN model and SSD model to validate its feasibility and accuracy further and provide a new approach for the detection of coal and gangue. [ABSTRACT FROM AUTHOR]

The Transformer has become pivotal for the integrated analysis of multi-source remote-sensing (RS) data in Earth observation, particularly in applications such as the fusion classification of hyperspectral images (HSI) and Light Detection and Ranging (LiDAR) data. However, Transformers are often employed as effective feature extractors by adopting similar processing blocks for different modalities from multi-source sensors, overlooking differences in imaging principles and data characteristics. Moreover, in the feature extraction process across different sensor data, there is a lack of necessary cross-modal information interaction, leading to insufficient utilization of complementary information between different sensors and resulting in suboptimal fusion outcomes. In this paper, we propose an interactive Transformer and CNN network for the fusion classification of HSI and LiDAR data. Specifically, a heterogeneous three-branch network architecture is designed for HSI and LiDAR data, where Transformers and CNNs encapsulate global contextual spatial and spectral information for HSI and capture geometric elevation patterns for LiDAR data, respectively. Elevation-Spatial Interaction (ESI) and Spectral-Spatial Interaction (SSI) modules are then introduced for multi-stage feature interaction. ESI enables the CNN-Transformer network to focus on essential local elevation details while simultaneously modelling global contextual spatial information. SSI facilitates the Transformer-Transformer network to cyclically intertwine spectral and spatial information for long-range spectral-spatial feature fusion. Finally, the interacted elevation, spatial, and spectral features undergo the Gated Fusion module to achieve hierarchical fusion adaptively, resulting in an elevation-spatial-spectral representation. Experiments conducted on three benchmark HSI-LiDAR datasets demonstrate the effectiveness of our proposed approach. [ABSTRACT FROM AUTHOR]

As a deep learning method, Convolutional Neural Network (CNN) can be used in image recognition, fault diagnosis and so on. Due to the internal parameter optimisation problem, the Particle Swarm Optimisation (PSO) has been introduced, but PSO is easy to fall into local optimal solution. In this paper, an adaptive CNN based on Quantum Particle Swarm Optimisation (QPSO-CNN) is proposed and applied to rolling bearings fault diagnosis, which increases the richness of particles and makes it easy to find the global optimal solution. Firstly, the one-dimensional time-series data is compressed by piecewise aggregate approximation algorithm and converted into the heat map by the Gramian angular field algorithm; Secondly, QPSO algorithm is used to search the best CNN model to fit the data set; Finally, the training and validation set are used to search the best network architecture, and the test set is used to test the diagnostic accuracy of the best CNN model, which show that the proposed method has high accuracy. [ABSTRACT FROM AUTHOR]

The paper proposes a damage imaging identification method for plate structures based on piezoelectric Lamb waves and the MUSIC algorithm, focusing on enhancing the accuracy of damage identification by using the sensing signal mode separation technique, and on the MUSIC-algorithm-based damage level evaluation. The mode separation of damage scattering signals is realised by using pairwise geometric symmetry to simplify the processing of sensing signals and improving the accuracy of damage imaging. A damage factor constructed by using the maximum eigenvalue of the sensing signal matrix is proposed, and it is combined with imaging technology to realise the precise locating and preliminary evaluation of damage. A combination of finite element simulation and experimental validation is used to verify the effectiveness of the proposed method. The results show that accurate damage localisation imaging can be obtained by using a mode-separated damage scattering signal; the damage level has an approximately linear relationship with the damage factor, which can be used to initially recognise the damage level. Within a certain level of damage, the proposed method has high accuracy and stability, enhancing the effectiveness of the damage identification technology for plate structures based on the MUSIC algorithm. [ABSTRACT FROM AUTHOR]

Feature extraction is a core aspect in hyperspectral image classification, which can extract key information closely related to ground cover from complex scene, thus improving classification accuracy. Therefore, designing an effective feature extraction network is a hotspot and a challenge in the current research. In this paper, a feature extraction framework based on 3-D homogeneous attribute decomposition (3D-HAD) is proposed for HSI classification, which consists of the following key technologies. First, the principal component analysis algorithm is applied to the raw HSI to extract the principal components (PCs), and the raw HSI is clustered into many 3D superpixel blocks according to the first three PCs-based over-segmentation strategy. Then, a superpixel intrinsic attribute decomposition (SIAD) is designed to capture reflectance feature and suppress shading feature. Meanwhile, a metric entropy is introduced into the decomposition process to overcome the spectral-spatial weak assumption among pixels. Next, superpixel-guided recursive filtering is employed to preserve global details of HSI to enhance accuracy in HSI classification. Finally, the support vector machine classifier is used to obtain classification results of HSI. Experiments performed on several real hyperspectral data sets with limited training samples indicate that the proposed 3D-HAD method outperforms the classic, advance, and deep learning classification methods. [ABSTRACT FROM AUTHOR]

Omni-dimensional dynamic convolution can adjust spatial sizes, the amount of input/output channels, and kernel size based on the characteristics of the input, thus improving the capacity to extract characteristics through convolution, resulting in increased computation and model complexity. As a result, it is challenging to deploy these models on resource-limited airborne high-spectral testing systems. This paper proposes a lightweight omni-dimensional dynamic convolution network as a possible solution to this issue. In this network, the LS2CM module is used for spatial-spectral information extraction, and the pyramidal residual network is the basic architecture for extracting features from shallow to deep layers, and lightweight omni-dimensional dynamic convolution is adopted in the convolution module. Overcoming the pyramidal structure can only extract local information, we proposed a new simple spatial-spectral self-attention mechanism to extract global information from hyperspectral. Besides, the Huber function as an activation function. The IP, WHU-Hi-LongKou, SA, and PU datasets were used to verify the accuracy of the suggested network model. The results show that the proposed lightweight network maintains an appropriate equilibrium among model complexity as well as accuracy. [ABSTRACT FROM AUTHOR]

This paper mainly addresses the lack of labelled data and insufficient data utilization in PolSAR image classification. We propose a channel adaptive Complex-Valued Fully Convolutional Networks based on Transfer learning (TF-CVFCN). To solve the problem of less labelled data, we borrowed from the transfer learning method, which can use a large amount of easily obtained relevant or even irrelevant data for pre-training and then use a small amount of source data for network training fine-tuning. Since PolSAR data can obtain different data information by using different decomposition methods, the current research trend is to process PolSAR data in the complex-valued containing more information. In our work, to cater to the proposed network model and make the network design more liberal, a new transfer learning method is proposed, which uses a new method to realize partial layer parameter transfer, and achieves the purpose of optimizing the traditional Transfer learning method. On this basis, we optimize the network structure to improve the classification accuracy of PolSAR data, and evaluate its performance on three datasets. Furthermore, we lighten the TF-CVFCN and optimize the relevant parameters using Competitive Multi-Objective Particle Swarm Optimizer, emphasizing the importance of this parameter optimization in large networks. [ABSTRACT FROM AUTHOR]

Traditional methods for coal gangue sorting exhibit low efficiency, significant safety hazards, and limited applicability. Existing machine vision-based coal gangue image recognition methods struggle to balance model recognition speed and accuracy. In response to these challenges, this paper first utilizes the improved MSRCR algorithm to process images, enhancing the dark areas of coal gangue images while ensuring uniform brightness enhancement and image clarity. Furthermore, a novel lightweight coal gangue recognition method is proposed based on YOLOv8n, aiming to reduce data redundancy and improve recognition accuracy. Experimental results demonstrate that the improved lightweight model has a computational load of 7.1 GFLOPs, representing only 86.6% of the original model. The model detection rate is 73 fps, a 17 fps improvement over the original model. The accuracy, recall rate, and average precision reach 98.1%, 97.6%, and 98.9%, respectively, with improvements of 1.1, 0.9, and 0.5% points over the original model. The missing detection phenomenon is avoided effectively, and the accuracy and portability of the model are improved. [ABSTRACT FROM AUTHOR]

This paper aims to evaluate the performance of a hydrological model by using satellite image classification (SIM) to extract land use (LU) information. Four methods, namely naive Bayes (NB), classification and regression trees (CART), support vector machine (SVM), and random forest (RF), are assessed for SIM using satellite images. CART demonstrated the highest overall accuracy (OA) of 0.996, followed by RF (OA = 0.994), SVM (OA = 0.797), and NB (OA = 0.543). The Soil and Water Assessment Tool (SWAT) model was employed in this study to construct and refine the hydrological model for simulating the streamflow in the basin. For this purpose, the ground-based layer was replaced by LU generated from the classification algorithms. Integrating the four classification approaches yielded a significant improvement in Nash-Sutcliffe efficiency of the SWAT model, increasing it from 0.37 to approximately 0.76. These findings highlight the effectiveness of using satellite image classification in enhancing the efficiency of hydrological models. [ABSTRACT FROM AUTHOR]

Historical fine-scale information of archaeological landscapes is crucial in archaeological investigations. However, documenting such information using satellite sensor data prior to 2000 remains a daunting challenge. Images from the declassified archives of KH-9 HEXAGON (KH-9) cameras, including the panoramic camera system (PCS) and mapping camera system (MCS), offer fine-scale information about archaeological sites. However, noise, contrast distortion and the availability of only a single panchromatic band can limit their potential, particularly for identifying features in subtropical climates within heterogeneous landscape types. This paper focuses on developing a novel multifaceted analytical framework with two components: image pre-processing and feature identification. The image pre-processing component is divided into two steps. First, a trained stationary wavelet transform (SWT) based on the normalized sill (NS) is developed to not only de-noises the image, but also preserve its original image characteristics. Then, the contrast of the de-noised images is optimized by the multi-resolution Top-hat (MTH) using multi-scale information. In the feature identification component, the MCS image is analysed using spatial colour composite write function memory (SCCWFM) and spatial novelty detection (SND). An ultra-fine spatial three-dimensional colour composite (UFSTCC) image and ultra-fine spatial digital surface model (UFSDSM) are produced to aid interpretation of the KH-9 PCS images. The proposed processing pipelines were tested on KH-9 MCS and PCS images of the World Heritage site at Liangzhu Ancient City (LAC) in China, which is characterized by a subtropical climate and a heterogeneous landscape types. The proposed pre-processing pipeline improved considerably the appearance of these images across the LAC landscape while maintaining the original image information. The developed digital analytical approaches for KH-9 PCS and MCS images facilitated straightforward identification of archaeological features in the LAC. The proposed framework has the potential to increase exploitation of the available KH-9 images in archaeological applications. [ABSTRACT FROM AUTHOR]

The classification of very high-resolution satellite imagery remains a focal point in remote sensing, attracting increased attention across diverse scientific disciplines. Various classification methods, including pixel- and object-based techniques, have been proposed, and their performances and limitations have been discussed in the literature. This paper presents a hybrid method that combines the strengths of pixel- and object-based methods in image classification to minimize errors associated with the segmentation process, particularly under-segmentation errors in object-based image analysis. The core concept behind the method lies in categorizing segmented image objects as either homogeneous or heterogeneous based on their class probability. In this process, the estimated possibilities from the object-based classification model are considered, and segments are designated as homogeneous or heterogeneous using a user-defined threshold. The object-based classification model determines the class labels for homogeneous image objects, while the heterogeneous ones, containing pixels representing different land cover classes, are classified using the pixel-based model. The performance of hybrid classification models, created by varying thresholds, is analysed using high-resolution WorldView-3 and WorldView-2 imagery and compared with pixel- and object-based classification results. For the implementation of image classification methods, Canonical Correlation Forest (CCF), Extreme Gradient Boosting (XGBoost), and Categorical Boosting (CatBoost) were employed. The findings indicated that employing the suggested hybrid strategy with a threshold value selected within a specific range (e.g. between 60% and 80%) and employing a robust classification algorithm that provides class probabilities (e.g. CCF) results in a statistically significant improvement in overall accuracy compared to pixel and object-based methods, with gains of 5% and 4%, respectively. Visual analysis of the produced thematic maps revealed that the proposed method minimizes the salt-and-pepper effect associated with pixel-based classification while mitigating segmentation problems arising from object-based classification. [ABSTRACT FROM AUTHOR]

AbstractResearch in explainable AI (XAI) aims to provide insights into the decision-making process of opaque AI models. To date, most XAI methods offer one-off and static explanations, which cannot cater to the diverse backgrounds and understanding levels of users. With this paper, we investigate if free-form conversations can enhance users’ comprehension of static explanations in image classification, improve acceptance and trust in the explanation methods, and facilitate human-AI collaboration. We conduct a human-subject experiment with 120 participants. Half serve as the experimental group and engage in a conversation with a human expert regarding the static explanations, while the other half are in the control group and read the materials regarding static explanations independently. We measure the participants’ objective and self-reported comprehension, acceptance, and trust of static explanations. Results show that conversations significantly improve participants’ comprehension, acceptance , trust, and collaboration with static explanations, while reading the explanations independently does not have these effects and even decreases users’ acceptance of explanations. Our findings highlight the importance of customized model explanations in the format of free-form conversations and provide insights for the future design of conversational explanations. [ABSTRACT FROM AUTHOR]

This paper introduces Knowledge Distillation of Vision Transformer (KDViT), a novel approach for medical image classification. The Vision Transformer architecture incorporates a self-attention mechanism to autonomously learn image structure. The input medical image is segmented into patches and transformed into low-dimensional linear embeddings. Position information is integrated into each patch, and a learnable classification token is appended for classification, thereby preserving spatial relationships within the image. The output vectors are then fed into a Transformer encoder to extract both local and global features, leveraging the inherent attention mechanism for robust feature extraction across diverse medical imaging scenarios. Furthermore, knowledge distillation is employed to enhance performance by transferring insights from a large teacher model to a small student model. This approach reduces the computational requirements of the larger model and improves overall effectiveness. Integrating knowledge distillation with two Vision Transformer models not only showcases the novelty of the proposed solution for medical image classification but also enhances model interpretability, reduces computational complexity, and improves generalization capabilities. The proposed KDViT model achieved high accuracy rates of 98.39%, 88.57%, and 99.15% on the SARS-CoV-2-CT, COVID-CT, and iCTCF datasets respectively, surpassing the performance of other state-of-the-art methods. [ABSTRACT FROM AUTHOR]