ABSTRACT The long‐tail distribution is characterized by a large number of samples in a few categories (head classes) and a scarcity of samples in most categories (tail classes). This inherent class imbalance significantly degrades the performance of conventional classification models, particularly on tail classes. To tackle this challenge, we propose a Multi‐Expert Dynamic Gating and Feature Decoupling Classification Algorithm based on Uniform Enhanced Sampling. The proposed method integrates multi‐expert learning with data augmentation and enhances tail classes performance by jointly optimizing the loss function and the expert assignment network. Specifically, a uniform enhanced sampling strategy is introduced to augment tail classes samples and increase their sampling frequency through resampling. During the feature learning stage, the shared layers of a convolutional network extract general features, while multiple expert models are trained independently. A feature decoupling technique is employed to separate generic and class‐specific features. In addition, a binary gating mechanism is designed to dynamically assign experts while preventing over‐reliance on specific categories. Extensive experiments on three benchmark long‐tailed classification datasets:CIFAR10‐LT, CIFAR100‐LT, and ImageNet‐LT—demonstrate that our method consistently outperforms existing state‐of‐the‐art approaches. Ablation studies further confirm the effectiveness of the uniform enhanced sampling strategy and the joint optimization of multi‐expert learning, showing that our algorithm successfully balances the model's attention across head and tail classes, thereby improving overall classification performance.

Few-shot learning can find the latent structure information between the support and query data by the similarity metric of meta-learning to construct the discriminative model for recognizing the new categories with the less labeled samples. Most existing methods attempt to model the similarity relationships among samples within meta-tasks for achieving this goal. However, the relationship assessment among samples from distinct meta-tasks is challenging due to the differing metric relationships inherent to each respective meta-task. To address this issue, high-order structure-preserving graph neural network (HOSP-GNN) can deeply explore the rich samples structure of the different meta-tasks to predict the label of the queried data based on the graph. HOSP-GNN can mine high-order structures to enhance their relevance by triple samples. In addition, it can also generate the updating rule of the different-order structures for node and edge representation optimization under manifold constraints. Notably, HOSP-GNN eliminates the need for retraining the learning model to recognize new classes, thanks to its high generalization of high-order structure that ensures model adaptability. The experiments demonstrate that HOSP-GNN outperforms state-of-the-art methods in four benchmark datasets, as well as on a self-built dataset focusing on endangered animals. The available code ishttps://github.com/yangfeifei02/HOSP.

Abstract Cross‐domain action recognition is a challenging vision task due to the domain shift and the absence of labeled data in the target domain. With only labelled source domain and unlabelled target domain data during training, some existing methods rely on an adversarial framework to align the features from different domains to a common latent space. However, the existing adversarial‐based approaches have a major limitation of only attempting to perform the alignment from a holistic view, ignoring the underlying coherence of class structure across domains. A class structure‐aware adversarial loss (CSCAL) is presented to address this issue. The CSCAL incorporates the category information into the adversarial learning branch to capture the fine‐grained alignment of each class, effectively avoiding the false mixup of samples from different categories in the embedding space. Experiments on HMDB51, UCF101 and Olympic Sports datasets show significant improvement compared to the baseline. Code and trained model can be found at https://github.com/bregmangh/CSCAL .

The query image is usually a simple and single object in image retrieval, and the reference images in the database usually have many distractions. The precision of image retrieval can be greatly improved If the target regions in the database image are extracted during retrieval. So this paper proposes a Bow image retrieval method based on SSD target detection. First, the training gallery is manually annotated to record the location and size information. Second, the SSD target detection model is trained with the labeled training gallery to obtain the target object SSD model. Third, the SSD model is used to locate the similar target regions of the reference image and the query graph. Finally, the target region information is mapped into the convolutional features, and these feature vectors are used for image similarity matching. The performance of the proposed method is evaluated on Paris6k, Oxford5k, Paris106k and Oxford105k databases. The experimental results show that the accuracy of image retrieval will be greatly improved by adding optimization methods in the proposed image retrieval framework. The image retrieval accuracy of this method is higher than that of similar methods in recent years.

Effective detection of infrared (IR) moving small targets in complex cluttered environments plays a key role in IR search and track systems for self-defense or attacks. In this letter, an IR moving small-target detection algorithm utilizing a spatiotemporal consistency of motion trajectories is proposed. First, feature points are densely sampled and tracked using the dense optical flow algorithm to compute dense trajectories. Second, suspected trajectories are deleted by utilizing the moving characteristics of the target. Third, under the assumption that each small target is defined as a compact space region, a binary image is created depending on the image coordinates of the trajectory points, from which salient contours are extracted as candidate target regions. Finally, a coding mechanism for contour numbering is introduced, and the moving targets are distinguished from the backgrounds by the temporal consistency of contour codewords. Several experiments were conducted, and their results demonstrate that our proposed method can detect small moving IR targets with higher detection rate, lower false alarm rate, and less running time compared with the state-of-the-art methods.

Due to the explosive increase in online videos, near-duplicate video retrieval (NDVR) has attracted much researcher attention. NDVR has very wide applications, such as copyright protection, online video monitoring, and automatic video tagging. Local features serve as elementary building blocks in many NDVR algorithms, and most of them exploit the local volume information using a bag of features (BOF) representation. However, such representation ignores potentially valuable information about the global distribution of interest points. Moreover, the discriminative power of the local descriptors is significantly reduced by the quantizer in BOF. Our motivation is that if we use the global features to classify the same or similar keyframes into the same class, it will be very useful in improving the performance of NDVR. In this paper, we present an improved radon transform (IR) feature which captures the detailed global geometrical distribution of interest points. It is calculated by using the 2D discrete Radon transform, and then applying a principal component analysis. Such IR feature is not only invariant to the geometry transformations but also robust to the noises. In addition, we propose a fusion strategy to combine the BOF representation with the global IR feature for further improving the recognition accuracy. Convincing experimental results on several publicly available datasets demonstrate that our proposed approach outperforms the state-of-the-art approaches in NDVR.

Structure fusion (SF) has been presented for multiple feature fusion via mining the discriminative and complementary information from different feature sets. As the typical methods, SF based on locality preserving projections (SFLPP) and SF based on tensor subspace analysis (SFTSA) have been developed for classification by capturing the complete structure from different features. However, the jointed optimisation function of SFLPP or SFTSA does not clearly explain the modelling mechanism of SF, and its solving process is complex because of iterative eigenvalue decomposition. In this study, structure modelling based on maximisation posterior probability (SMMPP) is proposed for solving these issues. It jointly considers both the certain prior structure (the mutual structure of multiple feature structure described by Ising model) and the uncertain likelihood structure (the possible fusion structure of multiple feature structure represented by Markov random field model) into the framework of Bayes’ rule. The proposed computational solution is faster‐converging speed than SFLPP or SFTSA with the guarantee of convergence. Extensive experiments conducted on shape analysis and human action recognition demonstrate the superiority of SMMPP over the state of art methods.

The paper studies fuzzy neural network theory and establishes structures of fuzzy neural network for intelligent compaction control. For compaction performance of a roller, fuzzy neural network parameters are self-corrected by learning algorithms of compensatory fuzzy neural network. Fuzzy control rules table that was educed in practice is looked at as training samples of fuzzy neural network. Simulation results show that the fuzzy neural network controller has generalization ability in error bound.