Accurate medical image segmentation is essential for reliable diagnosis, surgical planning, and disease monitoring. Semi-supervised medical image segmentation offers great potential by exploiting abundant unlabeled data with limited annotations, but it is prone to confirmation bias. To overcome this, we propose Attention Inverted Feature Perturbation (AIFP), a novel method that adaptively inverts feature-level attention weights to generate perturbations. This strategy encourages diversity and maintains independence between networks within a co-training framework, thereby mitigating confirmation bias. Extensive experiments on four public benchmarks validate the effectiveness of AIFP. Specifically, our method achieves Dice scores of 89.90% on ACDC and 91.01% on LA using only 10% labeled data, and 84.58% on Pancreas-NIH and 82.58% on PROMISE12 using 20% labeled data. These results consistently outperform state-of-the-art semi-supervised approaches, highlighting the practical value of AIFP in advancing accurate and robust medical image segmentation. AIFP enables reliable segmentation with limited annotations, supporting critical tasks such as left atrium delineation, pancreas boundary identification, and prostate segmentation. By reducing annotation demands while ensuring robustness, it has the potential to accelerate the clinical adoption of artificial intelligence-driven imaging tools.

Semi-supervised learning (SSL) offers a promising solution to the challenge of learning from limited labeled data by leveraging the potential of unlabeled data, thus circumventing the need for costly labeling efforts. However, common SSL methods often encounter domain shifts in many real-world scenarios, where class distribution is imbalanced. In order to make machine learning more robust to imbalanced datasets, it is imperative to ensure that consistent representations are learned for each class, regardless of the amount of data available. Therefore, we propose a straightforward yet effective kernel function mapping strategy to align the representations of each class in an infinite-dimensional space. Specifically, we employ a Gaussian kernel function to map the representations of unlabeled data to the centroids of labeled data, enabling similarity comparisons in the infinite-dimensional space. In this way, we are able to refine the predicted pseudo-labels at the representation level. To better handle class imbalance, we note that it is common to obtain a high recall but low precision for the majority classes and a high precision but low recall for the minority classes. A selective strategy is adopted for predictions corrected for the majority classes while maintaining confidence in the pseudo-labels assigned to the minority classes. Extensive evaluations on various benchmarks and training settings validate the superior performance of the proposed method compared to the existing relevant state-of-the-art approaches.

To mitigate the impact of noisy labels, many methods prioritize simple samples with reliable labels, often overlooking the valuable information in more challenging samples. This study introduces SRODET, a novel semi-supervised remote sensing object detection model that leverages sample complexity to extract accurate pseudo-labeled knowledge. We employ a dual-branch structure (DBS) to generate reliable pseudo labels for auxiliary supervision, enhancing joint supervision to derive high-quality pseudo labels from low-confidence predictions. This approach reduces the risk of losing object instances due to low-confidence scores, particularly for extreme scales. Additionally, we introduce a pseudo-label training strategy based on sample difficulty, evaluating complexity through object uncertainty and angular information from remote sensing images. Our experimental results show that SRODET achieves state-of-the-art performance in semi-supervised remote sensing object detection across various settings in the DOTA-v1.5 and HRSC2016 benchmarks.

We introduce LCMatch, a novel semi-supervised scene classification framework designed to enhance the performance of remote sensing image classification. Our method improves upon the existing FixMatch framework by incorporating a hierarchical structure for pseudo-label generation. The framework consists of three key modules: Hierarchical Cross-Random Combination, Adaptive Weighting Mechanism, and Label Alignment. These modules work synergistically to generate high-quality pseudo-labels, refining model predictions, and adaptively balancing the contributions of labeled and unlabeled data during training. In addition, we conduct extensive experiments on three widely used remote sensing datasets including AID, UCMerced, and NWPU-RESISC45. Results demonstrate that LCMatch outperforms state-of-the-art semi-supervised learning methods in terms of classification accuracy. Specifically, LCMatch exhibits robust performance even with a very limited number of labeled samples, also effectively handling class imbalance and distinguishing challenging categories.

Correspondence pruning aims to find correct matches (inliers) from an initial set of putative correspondences, which is a fundamental task for many applications. The process of finding is challenging, given the varying inlier ratios between scenes/image pairs due to significant visual differences. However, the performance of the existing methods is usually limited by the problem of lacking visual cues (e.g., texture, illumination, structure) of scenes. In this paper, we propose a Visual-Spatial Fusion Transformer (VSFormer) to identify inliers and recover camera poses accurately. Firstly, we obtain highly abstract visual cues of a scene with the cross attention between local features of two-view images. Then, we model these visual cues and correspondences by a joint visual-spatial fusion module, simultaneously embedding visual cues into correspondences for pruning. Additionally, to mine the consistency of correspondences, we also design a novel module that combines the KNN-based graph and the transformer, effectively capturing both local and global contexts. Extensive experiments have demonstrated that the proposed VSFormer outperforms state-of-the-art methods on outdoor and indoor benchmarks. Our code is provided at the following repository: https://github.com/sugar-fly/VSFormer.

Recently, deep convolutional neural networks (DCNNs) have been widely adopted in scene classification for remote sensing (RS) images, yielding impressive results. One important assumption for training these models is that the image labels are free from errors. However, the mislabeling of certain images can severely degrade the performance of DCNN models. While existing methods offer some degree of mitigation, their effectiveness diminishes significantly in the presence of highly contaminated labels. This letter seeks to address this issue by proposing a robust scene classification model designed to learn from noisy labels. First, a semi-supervised approach is proposed, involving bootstrapping a model that uses a cleaned training set through coarse screening of potentially noisy labels. In particular, the model is trained using consistency regularization to enhance its robustness. Second, the remaining images with uncertain labels are further used to increase the number of images available for training. Here, dynamic thresholding is applied to detect and correct label errors. Lastly, the images and their labels from the two steps above are jointly used for final training. The proposed method is evaluated on three widely used RS datasets (AID, WHU-RS19, and UCMerced). The results demonstrate that the proposed approach outperforms competing methods, highlighting its efficacy in addressing label noise in RS datasets.

Summary The fast and robust segmentation of tongue images is a prerequisite to achieve automatic tongue diagnosis in traditional Chinese medicine. In order to assist tongue diagnosis in real‐life scenarios, an ideal tongue segmentation method would need to obtain the entire tongue body as well as its precise contours. However, the similar appearance among the tongue body, the coating, and the lips hinders the performance for most unsupervised learning methods that primarily utilize low‐level visual features. On the other hand, although the supervised deep convolutional neural networks (DCNNs) that typically depend on the widely used crossentropy loss can achieve better accuracy, they are very prone to segment image to multiple trivially areas. To address both of the above issues, we make an attempt to boost the segmentation performance of DCNNs with a novel auxiliary loss function that seeks to exploit large margin learning for end‐to‐end tongue segmentation models. Specifically, we first propose a loss function that involves interclass and intraclass costs to directly measure the distance among pixels that belong to different connected regions. Then, we explore the potential ability of this loss function as a regularization for different segmentation networks such as those with attention modules and the deeply supervised network. Finally, a theoretical analysis for this learning scheme is presented. Through experiments on challenging datasets, we show that the proposed approach can be easily integrated into state‐of‐the‐art networks to boost their performance at the tongue segmentation task without bells and whistles.

In recent years, online sentiment texts are generated by users in various domains and in different languages. Binary polarity classification (positive or negative) on business sentiment texts can help both companies and customers to evaluate products or services. Sometimes, the polarity of sentiment texts can be modified, making the polarity classification difficult. In sentiment analysis, such modification of polarity is termed as polarity shifting , which shifts the polarity of a sentiment clue (emotion, evaluation, etc.). It is well known that detection of polarity shifting can help improve sentiment analysis in texts. However, to detect polarity shifting in corpora is challenging: (1) polarity shifting is normally sparse in texts, making human annotation difficult; (2) corpora with dense polarity shifting are few; we may need polarity shifting patterns from various corpora. In this article, an approach is presented to extract polarity shifting patterns from any text corpus. For the first time, we proposed to select texts rich in polarity shifting by the idea of natural annotation , which is used to replace human annotation. With a sequence mining algorithm, the selected texts are used to generate polarity shifting pattern candidates, and then we rank them by C-value before human annotation. The approach is tested on different corpora and different languages. The results show that our approach can capture various types of polarity shifting patterns, and some patterns are unique to specific corpora. Therefore, for better performance, it is reasonable to construct polarity shifting patterns directly from the given corpus.

We present a context aware object detection method based on a retrieve-and-transform scene layout model. Given an input image, our approach first retrieves a coarse scene layout from a codebook of typical layout templates. In order to handle large layout variations, we use a variant of the spatial transformer network to transform and refine the retrieved layout, resulting in a set of interpretable and semantically meaningful feature maps of object locations and scales. The above steps are implemented as a Layout Transfer Network which we integrate into Faster RCNN to allow for joint reasoning of object detection and scene layout estimation. Extensive experiments on three public datasets verified that our approach provides consistent performance improvements to the state-of-the-art object detection baselines on a variety of challenging tasks in the traffic surveillance and the autonomous driving domains.

We propose a structured Hough voting method for detecting objects with heavy occlusion in indoor environments. First, we extend the Hough hypothesis space to include both object location and its visibility pattern, and design a new score function that accumulates votes for object detection and occlusion prediction. In addition, we explore the correlation between objects and their environment, building a depth-encoded object-context model based on RGB-D data. Particularly, we design a layered context representation and allow image patches from both objects and backgrounds voting for the object hypotheses. We demonstrate that using a data-driven 2.1D representation we can learn visual codebooks with better quality, and more interpretable detection results in terms of spatial relationship between objects and viewer. We test our algorithm on two challenging RGB-D datasets with significant occlusion and intraclass variation, and demonstrate the superior performance of our method.

We address the glass object localization problem with a RGB-D camera. Our approach uses a nonparametric, data-driven label transfer scheme for local glass boundary estimation. A weighted voting scheme based on a joint feature manifold is adopted to integrate depth and appearance cues, and we learn a distance metric on the depth-encoded feature manifold. Local boundary evidence is then integrated into a MRF framework for spatially coherent glass object detection and segmentation. The efficacy of our approach is verified on a challenging RGB-D glass dataset where we obtained a clear improvement over the state-of-the-art both in terms of accuracy and speed.