The goal of Few-Shot Segmentation (FSS) is to segment images of novel categories using few labeled examples. However, FSS tasks face challenges such as over-segmentation and lack of generalization issues. This paper addresses these challenges by employing a triple attention mechanism (TAM) and a hierarchical decoding transformer (HDT). Specifically, TAM is proposed to enhance the model’s ability to focus on spatial regions within query features that are semantically relevant to the target category. The HDT module then aggregates the enhanced query features with the support features in a decoupled manner, generating dense features with pixel-level semantic relevance, which improves the segmentation ability on novel classes. Additionally, considering that class-level labels inside image can provide weak supervision for the segmentation task, this paper introduces a contrastive language image pretraining (CLIP) based model to enhance the segmentation performance. The Grad-CAM mechanism is utilized to convert the class logit scores from CLIP into localization heatmaps, effectively leveraging the text label information to provide prior localization cues for the model. Extensive experiments conducted on the PASCAL-5i and COCO-20i datasets demonstrate state-of-the-art performance. The experimental results validate the effectiveness of the proposed method, significantly improving the generalization and segmentation performance of few-shot semantic segmentation models on novel categories.

Visible–infrared person re-identification (VI-ReID) aims to match pedestrians across heterogeneous visible and infrared modalities. Existing methods predominantly focus on learning modality-shared representations for cross-modality matching, which often suppress modality-specific discriminative cues. Some recent works attempt to introduce modality-specific features; however, their primary objective remains improving shared features for direct cross-modality matching, leaving the intrinsic potential of modality-specific information underexplored. In this paper, we propose a Specific–Shared Feature Inference (SSFI) framework that exploits modality-specific and modality-shared features in a fundamentally different manner. Instead of directly using modality-specific features for cross-modality matching, we leverage their strength in modeling intra-modality relationships. Specifically, a Specific–Shared Feature Extractor (SSFE) is designed to disentangle the two types of representations while enabling effective interaction between them. Furthermore, we introduce a Cross-Modality Similarity Inference (CSI) module, which utilizes modality-specific features to construct intra-modality affinity among gallery samples and propagates such relational information to refine cross-modality similarity estimation based on modality-shared features. Extensive experiments on standard VI-ReID benchmarks demonstrate that the proposed method consistently and significantly outperforms state-of-the-art approaches.

Fine-grained object detection, an essential task in remote sensing image interpretation, is particularly challenging due to the extreme inter-class similarity. Existing fine-grained detectors, with prevalent approaches local region refinement, accurate localization guidance, and multimodal signal supplementation, primarily focus on refining features within the foreground to improve detection accuracy. However, their effectiveness noticeably compromised when the original data is insufficient. We reconsider the potential factors affecting fine-grained detection accuracy and argue that the detection network is prone to learning a few specific finegrained patterns when trained on a limited data scale or low data diversity, leading to feature representations that struggle to generalize to unseen data. Motivated by this, we introduce a regularization perspective construct a one-stage fine-grained detector, specifically by designing a noise regularization branch for the data space and a semantic regularization branch for the feature space. The noise regularization branch facilitates learning more generalized fine-grained patterns by performing background random blending and foreground random concealing operations to construct richer data patterns, while the semantic regularization branch contributes to learning discriminative features within foregrounds by leveraging semantic correlations inherent in fine-grained categories for semantic calibration. Two fine-grained datasets for airplane detection are employed to evaluate detection performance. Extensive experimental results demonstrate that our proposed fine-grained detector consistently outperforms existing detectors while maintaining competitive inference speed and model complexity, thereby confirming that the regularization perspective provides an effective and practical approach for constructing fine-grained detectors with robust detection performance.

Video-based person re-identification (Re-ID) aims to identify video sequences of specific pedestrians captured in a distributed camera system. The key to tackling this challenging task is to model rich temporal features in video sequences. However, existing methods usually capture only fixed-length global temporal clues, making it difficult to eliminate the effect of occlusion and accurately model the motion information of pedestrians. To address this limitation, we propose a novel graph-based framework, namely dynamic hierarchical graph network(DHGN), to model temporal features by adaptively capturing multi-granularity temporal clues. Specifically, we adaptively segment the video features horizontally into several regions and construct graphs for the features at the same level. The features of each frame at the same level are considered as graph nodes, which are then adaptively connected based on feature similarity. By dynamically aggregating features from neighboring nodes in the same graph, DHGN is able to adaptively capture temporal cues from different body parts. Furthermore, in order to obtain more robust matching results, we propose a similarity weighted inference module (SWIM), which utilizes gallery-gallery similarity to modify the query-gallery similarity matrix. Extensive experiments on four benchmarks clearly demonstrate the effectiveness of the proposed method.

Taylor's formula holds significant importance in function representation, such as solving differential difference equations, ordinary differential equations, partial differential equations, and further promotes applications in visual perception, complex control, fluid mechanics, weather forecasting and thermodynamics. However, the Taylor's formula suffers from the curse of dimensionality and error propagation during derivative computation in discrete situations. In this paper, we propose a new discrete differential operator to estimate derivatives and to represent continuous smooth function locally using the Vandermonde coefficient matrix derived from truncated Taylor series. Our method simultaneously computes all derivatives of orders less than the number of sample points, inherently mitigating error propagation. Utilizing equidistant uniform sampling, it achieves high-order accuracy while alleviating the curse of dimensionality. We mathematically establish rigorous error bounds for both derivative estimation and function representation, demonstrating tighter bounds for lower-order derivatives. We extend our method to the two-dimensional case, enabling its use for multivariate derivative calculations. Experiments demonstrate the effectiveness and superiority of the proposed method compared to the finite forward difference method for derivative estimation and cubic spline and linear interpolation for function representation. Consequently, our technique offers broad applicability across domains such as vision representation, feature extraction, fluid mechanics, and cross-media imaging.

As object detection tasks progress rapidly, fine-grained detection flourishes as a promising extension. Fine-grained recognition naturally demands high-quality detail signals; however, existing fine-grained detectors, built upon the mainstream detection paradigm, struggle to simultaneously address the challenges of insufficient original signals and the loss of critical signals, resulting in inferior performance. We argue that language signals with advanced semantic knowledge can provide valuable information for fine-grained objects, as well as the frequency domain exhibits greater flexibility in suppressing and enhancing signals; then, we propose a fine-grained aircraft detector by integrating language knowledge and frequency representations into the one-stage detection paradigm. Concretely, by considering both original signals and deep feature signals, we develop three components, including an adaptive frequency augmentation branch (AFAB), a content-aware global features intensifier (CGFI), and a fine-grained text-image interactive feeder (FTIF), to facilitate perceiving and retaining critical signals throughout pivotal detection stages. The AFAB adaptively processes image patches according to their frequency characteristics in the Fourier domain, thus thoroughly mining critical visual content in the data space; the CGFI employs content-aware frequency filtering to enhance global features, allowing for generating an information-rich feature space; the FTIF introduces text knowledge to describe visual differences among fine-grained categories, conveying robust semantic priors from language signals to visual spaces via multimodal interaction for information supplement. Extensive experiments conducted on optical and SAR images demonstrate the superior performance of the proposed fine-grained detector, especially the FTIF, which can be plugged into most existing one-stage detectors to boost their fine-grained recognition performance significantly.

Ultrasonic nondestructive testing has been widely used in various industries due to its simple operation and harmlessness for the object to be detected. However, due to the mechanism of ultrasonic image generation, the generated ultrasonic images often have low resolution, which greatly affects the final detection results. How to improve the resolution of ultrasonic images has become the key to improving the accuracy of defect detection. Therefore, this paper proposes an ultrasonic super-resolution model based on up- and down-sampling layers and multi-layer residual networks combined with Charbonnier loss function. The degradation features of the image are learned through up- and down-sampling layers, and the intrinsic features of the image are learned through multi-layer residual networks, so that all the feature information of the image is fully learned. The Charbonnier loss function accelerates the convergence of the model. Experimental results show that the model proposed in this paper outperforms the common model performance.

In remote sensing image classification task, it is often apply a model trained on one dataset (source domain) to another dataset (target domain). However, due to the presence of domain shift between these domains where data are not independent and identically distributed, the performance of the model typically deteriorates. Domain adaptation aims to improve the generalization performance of the model in the target domain. In response to the challenges of intricate backgrounds, domain shift, and potentially unlabeled target domain in remote sensing images, this article proposes a network specifically designed for unsupervised domain adaptation (UDA) classification of remote sensing images, named PPLM-net. The network consists of a domain adversarial training (DAT) module, a partial patch local masking (PPLM) module and a teacher-student network module. The DAT module enables the network to extract domain-invariant features. The PPLM module compels the model to focus on the global information of target domain remote sensing images with intricate backgrounds, learning contextual content to improve model performance. The teacher network generates pseudolabels for complete unlabeled target domain images. The student network trained with PPLM target domain classification loss to generate robust and discriminative features. We construct a dataset dedicated to the UDA scene classification task of remote sensing images named RSDA. We collect images from four publicly available datasets spanning seven common categories, containing over 10 000 images. Compared with the current state-of-the-art UDA model, PPLM-net achieves the best results in 12 domain adaptation classification tasks on RSDA. The average accuracy reaches 99.115%.

Microsatellite instability (MSI), characterized by alterations in short tandem repeat sequences throughout the genome, is an important biomarker for cancer diagnosis, prognosis, and treatment planning. Traditional MSI detection methods, such as immunohistochemistry (IHC) and polymerase chain reaction (PCR), are expensive, time-consuming, and resource-intensive. Recent advances in deep learning have enabled direct MSI prediction from hematoxylin and eosin (H&E) stained whole slide images (WSIs), offering a potentially faster and more cost-effective approach. However, existing convolutional neural network (CNN)-based methods for MSI prediction often focus on patch-level prediction and therefore struggle to capture the complex, multi-scale nature of histological features and their spatial relationships. To address these limitations, we propose a novel three-level multiscale graph neural network (GNN) method that integrates tissue, patch, and cellular-level analyses within a unified framework, namely TPC-GNN. The TPC-GNN extracts comprehensive biological information from WSIs by simulating structural interactions across various scales. At the tissue level, it captures global morphological patterns and tissue organization. The patch level focuses on local texture and architectural features, while the cellular level analyzes individual cell morphology and distribution. This multi-scale approach enables the model to capture complex biomarkers from micro to macro levels, providing a more holistic representation of the histological image. To enhance the integration of multi-scale information, we investigate different learnable fusion schemes based on multi-layer perceptrons (MLP), Transformers, and Mamba architectures. Experimental results demonstrate the superiority of our approach, achieving performance improvements of over 2% and 4% in Area Under the Curve (AUC) on the CRC-MSI and STAD-MSI datasets, respectively, across different network backbones compared to existing methods. These findings not only validate the effectiveness of our multiscale GNN approach but also highlight its application to complex histopathological-level image analysis. Codes are available at https://github.com/zhillusion/TPC-GNN.

The Gaussian homotopy method is a classical optimization approach for solving nonconvex problems. It applies Gaussian smoothing to a given problem, using varying smoothing factors to generate a series of proxy problems that are easier to solve, subsequently addressing them gradually from simpler to more complex ones. Traditional Gaussian homotopy methods typically utilize a predefined sequence or parameter to update a single smoothing factor serially. However, this sequential updating process is inefficient and unsuitable for parallel homotopy pipelines. Moreover, the utilization of predefined sequences or parameters leads to a lack of adaptability. To address the abovementioned challenges, we propose an adaptive parallel Gaussian homotopy optimization method. Initially, we introduce an adaptive model called Scaling. This model is formulated as a scaler that concurrently adjusts multiple smoothing factors, with each factor aligning to a homotopy level. It can be integrated easily into serial and parallel homotopy pipelines. Furthermore, we establish a collaborative training regimen to jointly train the Scaling model and a parallel homotopy model, named continuation path learning (CPL) [1] model. Throughout the training process, the Scaling model furnishes CPL with multiple scaled smoothing factors and updates CPL implicitly. Extensive experiments demonstrate that the proposed Gaussian homotopy approach performs competitively.

Infrared images are widely used in security monitoring and autonomous driving due to their resistance to light changes and adverse weather conditions. But, the low contrast and colorless characteristics of infrared images limit the effectiveness of human observation and subsequent detection and recognition algorithms. Methods for translating infrared images into visible images can overcome the above shortcomings, among which contrastive learning methods using self-similarity features achieve the best performance. However, methods based on contrastive learning face the sub-optimization problem caused by random sampling. And the contrastive loss based on self-similarity features faces the problem of encoding entanglement when used for infrared-visible image translation, that is, the features extracted from different categories of regions cannot be distinguished. Therefore, we propose a cross-similarity guided contrastive learning method for infrared-visible image translation. First, to address the randomness and inefficiency of the contrastive loss random sampling process, a sampling strategy based on information entropy ranking of cross-similarity matrix is proposed to obtain sampling points for subsequent contrastive loss calculation. By calculating the information entropy of cross-similarity matrix between input and generated images and sorting them, the sampling points with the most information can be obtained. Second, to alleviate the encoding entanglement problem of the self-similarity contrastive loss due to the low contrast of infrared images, multi-scale spatially adjacent graph structure consistency loss and spatially separated graph structure consistency loss based on cross-similarity matrices are proposed. Experiments on KAIST and FLIR datasets show that the proposed method has the best score and visual performance compared with multiple advanced infrared-visible translation methods. Ablation experiments further illustrate the effectiveness of the method.

Multiple Object Tracking (MOT) which is an important research topic in computer vision, plays an important role in the fields of automatic driving and area monitoring. In the object dense scene, there is a phenomenon of occlusion between a large number of object. A large number of locally visible object lead to the degradation of the tracking performance of the multi-object tracking algorithm in this scene. In this paper, we propose a method that combines high-order modeling, future location feature revision, and conceptual features to address the missing object re-matching problem. Higher-order modeling enables more accurate approximations of actual functions. It modifies the current prediction with one or more features of future locations. The overall feature of one object is composed of multiple local conceptual features. The object can be expressed by the combination of several concept features when it is greater than a certain similarity threshold. Experimental results show that the above three optimization mechanisms can effectively alleviate the problems of multiobject tracking algorithms in dense object scenes, and the optimized algorithm has significantly improved accuracy in multiple tracking scenarios.

Establishing the relationships among hierarchical visual attributes of objects in the visual world is crucial for human cognition. The classic convolution neural network (CNN) can successfully extract hierarchical features but ignore the relationships among features, resulting in shortcomings compared to humans in areas like interpretability and domain generalization. Recently, algorithms have introduced feature relationships by external prior knowledge and special auxiliary modules, which have been proven to bring multiple improvements in many computer vision tasks. However, prior knowledge is often difficult to obtain, and auxiliary modules bring additional consumption of computing and storage resources, which limits the flexibility and practicality of the algorithm. In this paper, we aim to drive the CNN model to learn the relationships among hierarchical deep features without prior knowledge and consumption increasing, while enhancing the fundamental performance of some aspects. Firstly, the task of learning the relationships among hierarchical features in CNN is defined and three key problems related to this task are pointed out, including the quantitative metric of connection intensity, the threshold of useless connections, and the updating strategy of relation graph. Secondly, Relational Embedding Convolution (RE-Conv) layer is proposed for the representation of feature relationships in convolution layer, followed by a scheme called use & disuse strategy which aims to address the three problems of feature relation learning. Finally, the improvements brought by the proposed feature relation learning scheme have been demonstrated through numerous experiments, including interpretability, domain generalization, noise robustness, and inference efficiency. In particular, the proposed scheme outperforms many state-of-the-art methods in the domain generalization community and can be seamlessly integrated with existing methods for further improvement. Meanwhile, it maintains comparable precision to the original CNN model while reducing floating point operations (FLOPs) by approximately 50%.

Video-based person re-identification (Re-ID) aims to retrieve the target person from video sequences captured by a distributed camera system. It remains a challenging task due to the reasons such as occlusion and misalignment in the video. To address above problem, many methods are proposed to exploit multi-scale spatio-temporal features in videos. However, established methods typically assign the equal weights to temporal or spatial features at different scales, which significantly diminishes the distinct roles of each feature. In this paper, we propose a novel Multi-scale Feature Aggregation Network (MFANet) for video-based person Re-ID. Specifically, we propose two flexible modules, Multi-scale Temporal Feature Aggregation (MTFA) and Multi-scale Spatial Feature Aggregation (MSFA). These two modules first extract different scales of temporal (dynamic, static) and spatial (coarse and fine) features, and then adaptively assign weights to each feature according to the video sequence. Both of these lightweight modules can be incorporated with 3D Convolutional Neural Network to build our MFANet. Extensive experiments on four public benchmarks demonstrate that MTFA and MSFA improve the performance of baseline architectures, and our MFANet achieves the best performance compared to other state-of-the-art methods.

Ultrasonic nondestructive testing has been widely used in industry due to its various advantages. However. with the increasing demand for non-destructive testing accuracy, low-resolution ultrasonic images can easily lead to misidentification of defects. How to improve the resolution of ultrasonic images has become a key issue restricting development of the non-destructive testing industry. Therefore. this paper proposes a super-resolution model for ultrasonic NDT images consisting of up and down sampling layers and deep residual networks. The model learns the degraded features of the image through the up and down sampling layers, and learns the intrinsic features of the image through the deep residual network, so as to learn the feature information of the image more completely to complete the super-resolution task of the image. Experimental results on ultrasound image datasets also validate the effectiveness of the model, outperforming other models in peak signa-to-noise ratio (PSNR) and structural similarity (SSIM) metrics.

The fine-grained object detection, capable of identifying subcategories or types, is thriving in remote sensing scenes. In practice, most existing fine-grained detectors are derived from the two-stage R-CNN paradigm with intricate anchor boxes, focusing on refining features of region of interest (RoI) to boost performance, which often incurs a redundant process. In contrast, the one-stage, anchor-free paradigm possesses a simple yet effective pipeline, but its exploration in fine-grained detections is still far from sufficient. In this article, we propose a one-stage, anchor-free fine-grained detector for remote sensing aircraft recognition. We initially delve into predominant issues when extending the one-stage framework to conduct fine-grained detections, typified by severe interclass confusion and inferior performance in rare categories. Then, we design a fine-grained classification branch, including a region-to-region context distributor (R2CD), a class-aware decoupled focal loss (CDFL), and a cross-shaped sample space (CS3), to address these hindrances. Specifically, the R2CD flexibly integrates the sparse attention mechanism with mask prediction operations to conduct region-level content interactions separately within the foreground and background of feature maps, significantly alleviating the interclass confusion by enhancing subtle features; the CDFL employs dynamic modulation factors driven by optimization gradients to regulate loss contributions across categories while optimizing category-specific heatmaps, thus prioritizing rare categories with hard samples; the CS3 attains a preferable assignment strategy of positive and negative samples by incorporating structure prior, facilitating the capture of foreground features. Extensive experiments conducted on the MAR20 and FAIRPlane11 datasets demonstrate that our model excels at distinguishing fine-grained categories and is well-suited for performing fine-grained detection tasks.

In recent years, the field of thermal to visible image transformation has made great progress with Generative Adversarial Network(GAN). Since GANs own the ability to fit the signal frequency, these methods are limited to using more GANs or designing sub-tasks to improve the quality of the generated image. Due to the large difference in the information frequency of individual objects in the image, GANs pay less attention to low-frequency information, which leads to edge distortion of dynamic objects such as people and vehicles in the generated image. Edge distortion has become an important issue for practical applications We propose a new structure and two new loss functions to complete the task of thermal to visible image transformation. Our network takes one thermal image as input and generates a visible RGB image as output. First, we down-sample inputs for three times to get multi-scale inputs. Then we build outputs in multi-scale form small size to large size and fuse the multi-scale outputs. In train stage, our two new losses better direct the network to converge. In order to prove the effectiveness of the proposed thermal to visible image transformation algorithm, we use the KASIT-MPD dataset, which includes real paired thermal and visible images.

In recent years, deep learning-based methods have made a big splash in the field of single image super-resolution (SISR), and a lot of excellent research works on lightweight image super-resolution networks have emerged. This series of approaches point a way forward for lightweight SR network design, but they generally suffer from two problems: 1) slow convergence of the network 2) poor image reconstruction quality. In this paper, we firstly propose a simple and effective learning rate decay scheme without additional cost, and the new strategy can accelerate the convergence speed of the network, and the model obtained in the same training time is about 0.05 dB higher in PSNR test results. Secondly, we think about the performance bottleneck of RFDN and propose an enhanced full-channel attention (EFA) module with a mixture of channel and spatial attention, which can improve about 0.085 dB in PSNR compared to the CCA attention module. Finally, we propose a new feature distillation network based on the EFA module, called EFAFDN. With only 523k parameters, the EFAFDN(x2) is lighter than the RFDN with 534k parameters and has better SR performance than the RFDN-L with 626k parameters. Extensive experimental results show that EFAFDN performs slightly better than RLFN, the winning model of the main track of the NTIRE 2022 Efficient Super-Resolution Challenge.

Ultrasonic nondestructive testing is widely used in industry due to its wide detection range and high detection efficiency. However, for multi-layer stacked rubber materials, ultrasound is easily affected by noise interference, it is a challenge work to reduce or separate the noise via signal processing. Therefore, a new ultrasonic echo denoising method based on symplectic geometry mode decomposition is proposed. It can fully decompose the useful components and noise components in the signal, and has a good screening effect on the useful components and noise components. In order to verify the effective-ness and robustness of the method, we conduct experimental research on the model with simulated and measured signals. The analysis results show that our method outperforms other noise reduction methods.(c) 2023 Elsevier Ltd. All rights reserved.

Background Due to the lack of further studies on the influence of age factors on plantar fasciitis, this study evaluates the characteristic observation points of magnetic resonance imaging in various age cohorts of patients with plantar fasciitis to help diagnosis. Methods A retrospective analysis of 160 cases of plantar fasciitis patients and normal subjects (who have the disease unrelated to plantar fasciitis) who have undergone an MRI examination in our institution. The two groups were separately divided into young adult subjects (36 to 44 years old), middle age adult subjects (45 to 59 years old), and older adult subjects (60 to 79 years old). Data was gathered regarding plantar fascia thickness, the coronal length of the plantar fascia at the calcaneal origin, the signal intensity of plantar fascia and surrounding structures, and the presence or absence of plantar calcaneal spurs, all of which were assessed objectively by the investigators. Results There were statistical differences in the thickness of plantar fascia between two groups of three age cohorts (Older adult patients: 0.59 ± 0.09 cm; Middle age adult patients: 0.49 ± 0.09 cm; Young adult patients: 0.47 ± 0.05 cm) (all p < 0.001). In addition, there were also statistical differences in the high signal intensity changes of the plantar fascia and surrounding soft tissues between two groups of three age cohorts (all p < 0.001). In older adult plantar fasciitis patients, with regard to plantar calcaneal spur discovery, there was a statistical difference between the two groups (Chi-square = 12.799. df = 1. p < 0.001). Conclusion In plantar fasciitis cases where a diagnosis is difficult, abnormalities in the soft tissue surrounding the plantar fascia in patients of low age are noteworthy. In older adult patients, the discovery of plantar calcaneal spurs with abnormal thickening of plantar fascia deserves attention, and abnormal MRI findings are more manifest. But the final diagnosis should be based on the medical history. Level of Evidence Level 3.