Multi-modal salient object detection (MSOD), which integrates complementary modalities such as depth or thermal data, primarily faces two challenges: accurately preserving salient object details and effectively aligning cross-modal features. Recent advances in using Stable Diffusion to generate images with fine edge details have inspired researchers to reformulate MSOD as a conditional mask generation process guided by salient features, which has achieved excellent visual results. However, these approaches often overlook the high computational cost and large-scale architecture of Stable Diffusion, both of which render it unsuitable for real-world MSOD applications. Therefore, we propose SimpleDiffusion, the first lightweight and efficient conditional diffusion model for MSOD that does not rely on Stable Diffusion. Specifically, we propose an Adaptive Cross-Modal Fusion Conditional Network and a Latent Denoising Network to reduce the complexity of diffusion models. Furthermore, we design a Multi-modal Feature Rectification and Fusion Module to enhance the representational capacity of cross-modal salient features. Customized training and sampling strategies are also developed to improve inference efficiency and reduce erroneous object segmentations. Experiments on multiple MSOD datasets demonstrate that SimpleDiffusion reduces model size by over tenfold and improves inference speed by more than fivefold compared to other diffusion-based methods, while maintaining comparable or superior performance.

Currently, most deep trackers follow the Siamese paradigm, where target tracking is achieved through similarity matching between the template and search region features. This approach often processes each image pair independently, making it difficult to establish sufficient temporal correlation. This leads to poor performance of trackers when dealing with scenes where target appearance changes drastically, such as partial occlusion, rapid motion, scale variations, and target deformation. To address the above issues, this paper proposes a Transformer-based tracker that can utilize accurate and updated historical context information. Specifically, by comprehensively fusing information from the template frame, historical context information(including the target’s location and surrounding state information), and search frame, our method is able to more accurately locate the target and significantly enhance the tracker’s robustness and regression accuracy. Furthermore, this paper introduces a feature enhancement network to strengthen the extracted target features of the search region, thereby suppressing the impact of background and noise on tracking performance. Based on the above design, the proposed method is referred to as HCETrack, which effectively leverages historical context information and a feature enhancement network. Our method can achieve remarkable performance improvements on the LaSOT, LaSOText, and GOT-10k datasets.

ABSTRACT The main challenge in visible‐infrared person re‐identification (VI‐ReID), which involves matching images of target persons across different modalities, is the significant modality gap between visible and infrared images. Current approaches generally fall into two categories: network architectures that map features from different modalities into a shared feature space, and methods that focus on modality generation and transformation. However, these methods often fail to fully account for contextual relationships, and the generated modalities may lack semantic interpretability. To address these limitations, this paper proposes a mixed modality generator that aligns the visible and infrared modalities as closely as possible within a unified feature space. To effectively leverage multi‐layered information, we introduce a hierarchical feature aggregation module, which establishes connections between features across different layers. Comprehensive experiments on the SYSU‐MM01, RegDB, and LLCM datasets demonstrate that our method significantly outperforms existing state‐of‐the‐art approaches, setting a new benchmark for VI‐ReID performance.

Pedestrian intrusion detection is crucial to ensuring safe railway operation. Current pedestrian detection algorithms lack consideration for real-world railway scenarios, such as the reflective properties of screen doors and train windows, may mistakenly trigger pedestrian intrusion alerts. Scale variability and pedestrian overlap often lead to detection inaccuracy, making them inadequate for addressing the specific requirements of railway perimeter security. This letter introduces an innovative pedestrian detection algorithm that incorporates Mirror Translation Attention (MTA) and Feature Pooling Enhancement (FPE). MTA, including mirror flipping and offsetting the feature mapping, could significantly mitigate missed detection caused by reflective surfaces. Additionally, we introduce sparsity to the inputs of the self-attention, which significantly enhancing the model's inference speed. A multi-scale approach is adopted to accommodate the diversity in pedestrian sizes, while the FPE addresses occlusion issues across various scales. Compared to the advanced YOLOv8 model, the proposed method improves AP50 by 1.6% to 92.11% and reduces model parameters by 63.55% in our self-built railway pedestrian intrusion dataset.

Dictionary learning is an efficient knowledge representation method that can learn the essential features of data. Traditional dictionary learning methods are difficult to obtain nonlinear information when processing large-scale and high-dimensional datasets. While most dictionary learning algorithms are based on the assumption that the training data and test data have the same feature distribution, which is not always true in practical applications. To address the above problems, we propose the Kernel Fisher Dictionary Transfer Learning (KFDTL) algorithm. First, we map each sample to high-dimensional space through kernel mapping and use any dictionary learning algorithm to learn the essential features. Then, the feature-based transfer learning method is performed to predict the labels of the target samples. This method includes three main contributions: (1) KFDTL constructs a discriminative Fisher embedding model to make the same class samples have similar coding coefficients; (2) Based on the relationship between profiles and atoms, KFDTL constructs an adaptive model that adapts source domain samples to target domain samples; (3) The kernel method is used to efficiently solve nonlinear problems. Experiments on a large number of public image datasets have proved the effectiveness of the proposed method. The source code of the proposed method is available at https://github.com/zzfan3/KFDTL .

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

Fast-developing single-cell technologies create unprecedented opportunities to reveal cell heterogeneity and diversity. Accurate classification of single cells is a critical prerequisite for recovering the mechanisms of heterogeneity. However, the scRNA-seq profiles we obtained at present have high dimensionality, sparsity, and noise, which pose challenges for existing clustering methods in grouping cells that belong to the same subpopulation based on transcriptomic profiles. Although many computational methods have been proposed developing novel and effective computational methods to accurately identify cell types remains a considerable challenge. We present a new computational framework to identify cell types by integrating low-rank representation (LRR) and nonnegative matrix factorization (NMF); this framework is named NMFLRR. The LRR captures the global properties of original data by using nuclear norms, and a locality constrained graph regularization term is introduced to characterize the data's local geometric information. The similarity matrix and low-dimensional features of data can be simultaneously obtained by applying the alternating direction method of multipliers (ADMM) algorithm to handle each variable alternatively in an iterative way. We finally obtained the predicted cell types by using a spectral algorithm based on the optimized similarity matrix. Nine real scRNA-seq datasets were used to test the performance of NMFLRR and fifteen other competitive methods, and the accuracy and robustness of the simulation results suggest the NMFLRR is a promising algorithm for the classification of single cells. The simulation code is freely available at: https://github.com/wzhangwhu/NMFLRR_code.

The typical sparse representation for classification (SRC) can obtain desirable recognition result when the training samples in each class are sufficient. Nevertheless, if the training sample set is small scale, i.e., each class has a few training samples, even single sample, the traditional SRC cannot perform well. Although one of the variants of the traditional SRC, the extended SRC(ESRC), can effectively address the above small-scale training set (SSTS) problem, its computational efficiency is very low and consequently constrains the application of the ESRC algorithm. In order to improve the computational efficiency of the ESRC algorithm, we propose a new algorithm based on coordinate descent scheme in this work. Our proposed algorithm is referred as to the fast extended SRC (FESRC) algorithm. Experiments on popular face datasets show that the FESRC algorithm can obtain the high computational efficiency without significantly degrading the recognition results.

Sparse representation for classification (SRC) has attracted much attention in recent years. It usually performs well under the following assumptions. The first assumption is that each class has sufficient training samples. In other words, SRC is not good at dealing with the undersampled problem, i.e., each class has few training samples, even single sample. The second one is that the sample vectors belonging to different classes should not distribute on the same vector direction. However, the above two assumptions are not always satisfied in real-world problems. In this paper, we propose a novel SRC based algorithm, i.e., kernel sparse representation based classifier for undersampled problem (KSRC-UP) to perform classification. It does not need the above assumptions in principle. KSRC-UP can deal well with the small scale and high dimensional real world data sets. Experiments on the popular face databases show that our KSRC-UP method can perform better than other SRC methods.

We modify the conventional principal component analysis (PCA) and propose a novel subspace learning framework, modified PCA (MPCA), using multiple similarity measurements. MPCA computes three similarity matrices exploiting the similarity measurements: 1) mutual information; 2) angle information; and 3) Gaussian kernel similarity. We employ the eigenvectors of similarity matrices to produce new subspaces, referred to as similarity subspaces. A new integrated similarity subspace is then generated using a novel feature selection approach. This approach needs to construct a kind of vector set, termed weak machine cell (WMC), which contains an appropriate number of the eigenvectors spanning the similarity subspaces. Combining the wrapper method and the forward selection scheme, MPCA selects a WMC at a time that has a powerful discriminative capability to classify samples. MPCA is very suitable for the application scenarios in which the number of the training samples is less than the data dimensionality. MPCA outperforms the other state-of-the-art PCA-based methods in terms of both classification accuracy and clustering result. In addition, MPCA can be applied to face image reconstruction. MPCA can use other types of similarity measurements. Extensive experiments on many popular real-world data sets, such as face databases, show that MPCA achieves desirable classification results, as well as has a powerful capability to represent data.

The linear discriminant analysis (LDA) is a very popular linear feature extraction approach. The algorithms of LDA usually perform well under the following two assumptions. The first assumption is that the global data structure is consistent with the local data structure. The second assumption is that the input data classes are Gaussian distributions. However, in real-world applications, these assumptions are not always satisfied. In this paper, we propose an improved LDA framework, the local LDA (LLDA), which can perform well without needing to satisfy the above two assumptions. Our LLDA framework can effectively capture the local structure of samples. According to different types of local data structure, our LLDA framework incorporates several different forms of linear feature extraction approaches, such as the classical LDA and principal component analysis. The proposed framework includes two LLDA algorithms: a vector-based LLDA algorithm and a matrix-based LLDA (MLLDA) algorithm. MLLDA is directly applicable to image recognition, such as face recognition. Our algorithms need to train only a small portion of the whole training set before testing a sample. They are suitable for learning large-scale databases especially when the input data dimensions are very high and can achieve high classification accuracy. Extensive experiments show that the proposed algorithms can obtain good classification results.