Discrimination-Aware Mechanism for Fine-grained Representation Learning

The main requisite for fine-grained recognition task is to focus on subtle discriminative details that make the subordinate classes different from each other. We note that existing methods implicitly address this requirement and leave it to a data-driven pipeline to figure out what makes a subordinate class different from the others. This results in two major limitations: First, the network focuses on the most obvious distinctions between classes and overlooks more subtle inter-class variations. Second, the chance of misclassifying a given sample in any of the negative classes is considered equal, while in fact, confusions generally occur among only the most similar classes. Here, we propose to explicitly force the network to find the subtle differences among closely related classes. In this pursuit, we introduce two key novelties that can be easily plugged into existing end-to-end deep learning pipelines. On one hand, we introduce “diversification block” which masks the most salient features for an input to force the network to use more subtle cues for its correct classification. Concurrently, we introduce a “gradient-boosting” loss function that focuses only on the confusing classes for each sample and therefore moves swiftly along the direction on the loss surface that seeks to resolve these ambiguities. The synergy between these two blocks helps the network to learn more effective feature representations. Comprehensive experiments are performed on five challenging datasets. Our approach outperforms existing methods using similar experimental setting on all five datasets.

In recent years, deep learning techniques have demonstrated their powerful feature extraction capabilities in a variety of visual analysis tasks. Especially in the field of fine-grained image classification, deep feature extraction technology has achieved great performance improvement. The depth features extracted based on the deep learning model have strong expressive power. Fine-grained image classification is intended to subclass large categories, such as different types of birds. This identification task is more challenging because these sub-classes have large interclass differences and small inter-class differences. This paper first introduces the research progress of fine-grained image classification. Then, the data set used in the experiment and the extended data set are introduced. A variable convolution method for fine-grained image classification is proposed, and the research direction of the next step is summarized and explored.

Automatic recognition of ground-based clouds is crucial for meteorology and especially for the operational safety of Unmanned Aerial Vehicles (UAVs), but it is challenged by variable cloud shapes, complex lighting, and background interference. This paper introduces ALGA-DenseNet, an improved DenseNet model with a multi-scale attention mechanism. The model employs Color Jitter to enhance image robustness and improve learning of intra-class variations and inter-class differences. It incorporates Adaptive Local and Global Attention (ALGA) to merge features, enhancing feature selection. Additionally, it integrates mixed and depthwise separable convolutions to optimize multi-scale feature extraction, reducing parameters and computational complexity. Furthermore, integrating a Vision Transformer (ViT) and Dynamic Multi-head Attention (DMA) enhances representation of complex cloud features. Experimental results show recognition accuracies of 97.94% on the TJNU (Tianjin Normal University) Ground-based Cloud Dataset (GCD) and 97.25% on the Cirrus Cumulus Stratus Nimbus (CCSN) dataset. This indicates the model’s capability for fine-grained, multi-scale extraction of cloud textures, shapes, and color features, along with strong generalization performance.

Introduction: Virtual humans have expanded the training opportunities available to healthcare learners. Particularly, virtual humans have allowed simulation to display visual cues that were not previously possible in other forms of healthcare training. However, the effect of virtual human fidelity on the perception of visual cues is unclear. Therefore, we explore the effect of virtual human rendering style on the perceptions of visual cues in a healthcare context.Methods: To explore the effect of rendering style on visual cues, we created a virtual human interface that allows users to interact with virtual humans that feature different rendering styles. We performed a mixed design user study that had undergraduate healthcare students (n = 107) interact with a virtual patient. The interaction featured a patient experiencing an allergic reaction and required trainees to identify visual cues (patient symptoms). The rendering styles explored include a 3D modeled virtual human and an AI generated photorealistic virtual human. Visual cues were applied using a Snapchat Lens.Results: When users are given a frame of reference (users could directly compare symptoms on both rendering styles), they rated the realism and severity of the photorealistic virtual human’s symptoms significantly higher than the realism of the 3D virtual human’s symptoms. However, we were unable to find significant differences in symptom realism and severity ratings when users were not given a frame of reference (users only interacted with one style of virtual humans). Additionally, we were unable to find significant differences in user interpersonal communication behaviors between the 3D and photorealistic rendering styles.Conclusion: Our findings suggest 1) higher fidelity rendering styles may be preferred if the learning objectives of a simulation require observing subtle visual cues on virtual humans and 2) the realism of virtual human rendering style does not necessarily affect participants’ interpersonal communication behaviors (time spent, questions asked).

Gait recognition via GEI subspace projections and collaborative representation classification

Performance comparison of feature selection and extraction methods with random instance selection

Ship object detection and fine-grained recognition of remote sensing images are hot topics in remote sensing image processing, with applications in fishing vessel operation command, merchant ship navigation route planning, and other fields. In order to improve the detection accuracy for different types of remote sensing ship objects, this paper proposes a ship object perception and feature refinement method based on the improved ReDet, called Mamba-ReDet (M-ReDet). First, this paper designs a ship object fine-grained feature extraction backbone (Mamba-ReResNet, M-ReResNet), which selects and reconstructs the unique features of different types of ship objects through the Mamba's selective memory to improve the algorithm's ability to extract fine-grained features. Secondly, the M-ReDet consists of the Ship Object Perception Module (SOPM) and the Ship Feature Refinement Module (SFRM), which can extract the ship's spatial position information from the feature map, fuse different scales of spatial position information and use this information to refine the fine-grained features to improve the detection accuracy of the algorithm for different categories of ships. Finally, we use the KFIoU and Focal Loss as the regression loss and classification loss of the algorithm to improve the accuracy of the training. The experimental results show that the mAP0.5 of the M-ReDet algorithm on the FAIR1M(ship) and DOTAv1.0 visible light (RGB) remote sensing image datasets are 43.29% and 82.09%, respectively, which is 2.78% and 3.34% higher than that of the ReDet.

Recent advancements in spatial transcriptomics technology have enabled the capture of gene expression profiles while maintaining spatial information. Accurately identifying spatial clustering plays a pivotal role in analyzing spatial transcriptomics data and understanding tissue microenvironments. However, current spatial domain identification methods cannot explore the complex relationship of gene expression profiles and spatial topology. To alleviate this issue, we propose STMCCL, a novel self-supervised learning framework that jointly trains a masked autoencoder and cluster-guided contrastive learning. This framework extracts informative latent representations from gene expression profiles and spatial information. Specifically, we first use data augmentation strategies to build augmented views and employ a masked encoder to generate a feature view. Then, encoders are applied to learn view-unique embeddings of each view. Furthermore, we introduce a multiple cluster-perspectives module that considers both geometric and structural relationships between clusters to produce more reliable cluster assignments. Finally, to derive more discriminative positives and negatives, the cluster-guided contrastive module calculates the confidence of each sample based on the initial cluster. Comprehensive experiments on 7 public datasets demonstrate that STMCCL outperforms the state-of-the-art baselines with finer-scale spatial domain identification.

Person re-identification with part prediction alignment

Autonomous driving perception systems are vulnerable to physical adversarial attacks. Existing defenses largely adopt loosely coupled architectures where visual and kinematic cues are processed in isolation, thus failing to exploit physical spatiotemporal consistency as a structural prior and often struggling to balance adversarial robustness, transferability, accuracy, and efficiency under realistic attacks. We propose a physics-aware trajectory–appearance consistency defense that detects and corrects spatiotemporal inconsistencies by tightly coupling visual semantics with physical dynamics. The module combines a dual-stream spatiotemporal encoder with endogenous feature orchestration and a frequency-domain kinematic embedding, turning tracking artifacts that are usually discarded as noise into discriminative cues. These inconsistencies are quantified by a Trajectory–Appearance Mutual Exclusion (TAME) energy, which supports a physics-aware switching rule to override flawed visual predictions. Operating on detector backbone features, outputs, and tracking states, the defense can be attached as a plug-in module behind diverse object detectors. Experiments on nuScenes, KITTI, and BDD100K show that the proposed defense substantially improves robustness against diverse categories of attacks: on nuScenes, it improves Correction Accuracy (CA) from 86.5% to 92.1% while reducing the computational overhead from 42 ms to 19 ms. Furthermore, the proposed defense maintains over 71.0% CA when transferred to unseen detectors and sustaining 72.4% CA under adaptive attackers.

This paper introduces a method for depression detection by applying feature extraction techniques to facial and cephalic data from the FacePsy dataset, as part of the ABC2025 Challenge on BeyondSmile: Detecting Depression through Facial Behavior and Head Gestures. Two feature extraction schemes—TSFEL and Catch22—were employed to derive features from Facial Action Units, Euler angles, eye-opening and smiling probabilities, and facial distances and angles computed from landmarks. Moreover, approximate entropy was extracted as a feature. Following feature extraction, feature selection and classification were performed using six machine learning models. In the Leave-One-Patient-Out cross-validation experiment, the Random Forest classifier achieved the best performance with an AUC of 1.00 using 100 combined features, surpassing the original work’s AUC of 0.67. For nested cross-validation with Leave-One-Day-Out and multimodality, the k -NN algorithm with SMOTE obtained an AUC of 0.94, outperforming the original result of 0.81. To enhance clinical interpretability, SHapley Additive explanations values were used to explain model decisions. The proposed approach demonstrates significant promise for depression detection in naturalistic settings using visual cues.

In the past three years, the stereo matching method based on 3D CNNs has achieved surprising results and has received more and more attention. However, most stereo matching approaches aim to improve prediction accuracy by constructing and aggregating cost volumes through extensive 3D convolutions, which not only does not fully utilize the geometric information, but also overlooks the computational speed. Thus, achieving high-accuracy, high-efficiency stereo matching has become challenging. In this paper, we present a rapid and precise stereo matching network named CGW based on 3D CNNs, which simultaneously achieves real-time functioning, considerable accuracy, and a strong generalization capability. The network is divided into two parts. The first part constructs the geometric attention cube through a lightweight feature extraction network and a lightweight 3D regularization network. The second part filters the context features using the geometric attention cube to obtain the context geometric cube, and finally, the disparity is predicted and refined to obtain the final disparity. We adopted MobileNetV3 as an efficient backbone for feature extraction and designed 3D depthwise separable convolutions with residual structures to replace traditional 3D convolutions for constructing the cost volume and performing cost aggregation, aiming to reduce the model size and improve the computational speed. Additionally, we designed the context geometric attention (CGA) module and embedded it into the lightweight 3D regularization network, as well as designed the Warped Disparity Refinement (WDR) network to further improve the disparity prediction accuracy. CGA effectively guides cost aggregation by integrating rich contextual and geometric information, while also providing feedback for feature learning to guide more efficient context feature extraction. WDR constructs a warping cost volume using the obtained initial disparity, combined with image features, the initial disparity map, and reconstruction errors, to optimize the disparity. According to the initial disparity, it searches for the accurate disparity within a refined range. By narrowing the search range, WDR simplifies the task for the network to locate the correct disparity (residual), while simultaneously improving the computational efficiency. Experiments conducted on multiple benchmark datasets showed that, compared to other fast methods, CGW has advantages in both speed and accuracy and exhibits better generalization performance.

PurposeThis paper aims to estimate contact location from sparse and high-dimensional soft tactile array sensor data using the tactile image. The authors used three feature extraction methods: handcrafted features, convolutional features and autoencoder features. Subsequently, these features were mapped to contact locations through a contact location regression network. Finally, the network performance was evaluated using spherical fittings of three different radii to further determine the optimal feature extraction method.Design/methodology/approachThis paper aims to estimate contact location from sparse and high-dimensional soft tactile array sensor data using the tactile image.FindingsThis research indicates that data collected by probes can be used for contact localization. Introducing a batch normalization layer after the feature extraction stage significantly enhances the model’s generalization performance. Through qualitative and quantitative analyses, the authors conclude that convolutional methods can more accurately estimate contact locations.Originality/valueThe paper provides both qualitative and quantitative analyses of the performance of three contact localization methods across different datasets. To address the challenge of obtaining accurate contact locations in quantitative analysis, an indirect measurement metric is proposed.

In the traditional person re-identification model, the CNN network is usually used for feature extraction. When converting the feature map into a feature vector, a large number of convolution operations are used to reduce the size of the feature map. In CNN, since the receptive field of the latter layer is obtained by convolution operation on the feature map of the previous layer, the size of this local receptive field is limited, and the computational cost is large. For these problems, combined with the self-attention characteristics of Transformer, an end-to-end person re-identification model (twinsReID) is designed that integrates feature information between levels in this article. For Transformer, the output of each layer is the correlation between its previous layer and other elements. This operation is equivalent to the global receptive field because each element needs to calculate the correlation with other elements, and the calculation is simple, so its cost is small. From these perspectives, Transformer has certain advantages over CNN's convolution operation. This paper uses Twins-SVT Transformer to replace the CNN network, combines the features extracted from the two different stages and divides them into two branches. First, convolve the feature map to obtain a fine-grained feature map, perform global adaptive average pooling on the second branch to obtain the feature vector. Then divide the feature map level into two sections, perform global adaptive average pooling on each. These three feature vectors are obtained and sent to the Triplet Loss respectively. After sending the feature vectors to the fully connected layer, the output is input to the Cross-Entropy Loss and Center-Loss. The model is verified On the Market-1501 dataset in the experiments. The mAP/rank1 index reaches 85.4%/93.7%, and reaches 93.6%/94.9% after reranking. The statistics of the parameters show that the parameters of the model are less than those of the traditional CNN model.

Breast imaging technique called mammography has gained bigger attention among the researchers for the diagnosis of breast malignancy in the woman. Mammogram screening is the most effective procedure to visualize various potential problems in the breast. The two most common features connected with breast tumors are mass lesions and microcalcification. The collection of suitable image preprocessing, segmentation, feature extraction, selection and prediction algorithms play an essential role in the accurate detection and classification of cancer on mammograms. Classification techniques estimate unlabeled datasets class labeling depending on its similarity to the pattern learned. The Glowworm Swarm Optimization(GSO) algorithm is ideal for finding several solutions, and dissimilar or equivalent objective function values at the same time. This feature of GSO is useful for optimizing the feature set obtained from multiscale feature extraction procedures. Poor performance in generalization is the issue that arises due to the unconditioned output matrix of the hidden stage of the ELM classifier. The optimization algorithms will address this matter because of their global search capabilities. This article suggests ELM with the Fruitfly Optimization Algorithm (ELM‐FOA) along with GSO to regulate the input weight to achieve maximal performance at the hidden node of the ELM. The testing precision and sensitivity of GSO‐ELM‐FOA are 100% and 97.91%, respectively. The system developed will detect the calcifications and tumors with an accuracy of 99.15%.