Sequence recognition models under a fully supervised paradigm require large-scale training data, incurring substantial annotation costs. Pseudo-labeling is one of the most effective techniques in semi-supervised learning, which leverages predicted confidence to filter pseudo-labels on unlabeled data for model training. However, recent studies indicate that the performance of semi-supervised learning is compromised by overconfident models, as the predicted unreliable confidences will filter noisy samples into training. In this work, we discover that the overconfidence in sequence recognition models is influenced by the linguistic properties of a sequence, where the tail character classes are prone to be mispredicted as the head ones that frequently appear in the language with high confidence. And this overconfidence continuously intensifies throughout the semi-supervised training process. To address this limitation, we propose a Dynamic Sequential Class-Aware Smoothing (DSCS) method that calibrates the overconfidence of the head class to alleviate the inaccurate pseudo-labeling caused by overconfident misprediction to improve the quality of pseudo-labels. Specifically, we design a sequential class-aware smoothing module that incorporates token class frequency information to regularize the model and prevent it from becoming overconfident toward the head class. Meanwhile, to address the overconfidence problem intensifying throughout the semi-supervised learning processes, we introduce a dynamic regularization module to adjust the calibration strength dynamically for the coordination between the calibration and semi-supervised learning processes. Extensive experiments demonstrate the effectiveness and generality of our method, which significantly reduces annotation efforts while maintaining competitive recognition performance.

Object detection in remote sensing imagery remains challenging due to extreme scale variations, and deployment computational inefficiency. Existing transformer-based detectors exhibit incomplete feature representations stemming from limited scale-awareness in handling multi-scale objects, insufficient modeling of negative-polarity interactions and uniform attention distributions. To address these limitations, this paper proposes a novel framework integrating a Global Edge Information Extraction (GEIE) module and Polarity-aware Linear Attention-based Internal Feature Interaction (PLAIFI) module. The GEIE module consists of two components to integrate multi-scale edge features: a Poly-Resolution Edge Feature Producer (PREFP) module and an Edge Fusion module. The PREFP module extracts edge information using a 3D SobelConv branch applied to shallow features and generates a multi-scale feature map to augment the hybrid encoder’s multi-scale representations with edge-aware priors. The Edge Fusion module supports flexible integration to enrich boundary-aware representations. To establish complementary feature enhancement, the PLAIFI module decomposes query-key pairs into positive/negative polarity components via learnable projection matrices, explicitly models same- and opposite-signed interactions through dual-polarity branches with polarity-sensitive coefficient matrices, and employs channel-wise learnable power functions to sharpen attention distributions by reducing entropy - enhancing scale-awareness while maintaining linear complexity. Extensive experiments demonstrate state-of-the-art performance: 95.0% mAP (+4.6%) on HRSC2016 (small-scale ships), 73.0% mAP (+5.9%) on DOTAv1.5 (multi-class objects), and 93.9% mAP on SIRSTv2 (tiny-scale infrared targets), achieved with 11.4% GFLOPs/11% Params increase over baseline RT-DETR. The principal innovation constitutes a synergistic framework where edge-aware feature recalibration and polarity-driven attention refinement mutually reinforce each other, establishing new state-of-the-art performance while maintaining deployment-friendly computational efficiency on satellite imagery. The source code will be available at https://github.com/Keykeykeykeykeykey/Global_Edge_Guidance.

In the emerging cloud-edge-end computing networks, edge servers possess more constrained resources and face greater task offloading pressure than centralized cloud servers due to the surge in mobile applications and data. Concurrently, the presence of multiple edge service providers introduces additional challenges, including competition among servers, disordered resource pricing, and a lack of coordination in edge and cloud resource allocation. To address these issues, we propose a novel approach aimed at optimizing task deployment, resource pricing, and system coordination. First, we develop a competitiveness model to facilitate efficient edge-side task allocation while addressing the challenges of resource pricing under competitive conditions. Second, we design a transformer-based task prediction model to enhance the accuracy of resource demand forecasting, thereby enabling more effective edge-cloud resource allocation. To achieve these objectives, the system's interaction is structured into two distinct stages. This division simplifies the problem-solving process and ensures that the long-term goal of maximizing benefits for all stakeholders—edge service providers, cloud providers, and end-users—is achieved. The proposed solution not only improves task offloading efficiency and resource utilization, but also promotes fair competition and pricing transparency across the system.

Deep sequence recognition (DSR) models receive increasing attention due to their superior application to various applications. Most DSR models use merely the target sequences as supervision without considering other related sequences, leading to over-confidence in their predictions. The DSR models trained with label smoothing regularize labels by equally and independently smoothing each token, reallocating a small value to other tokens for mitigating overconfidence. However, they do not consider tokens/sequences correlations that may provide more effective information to regularize training and thus lead to sub-optimal performance. In this work, we find tokens/sequences with high perception and semantic correlations with the target ones contain more correlated and effective information and thus facilitate more effective regularization. To this end, we propose a Perception and Semantic aware Sequence Regularization framework, which explore perceptively and semantically correlated tokens/sequences as regularization. Specifically, we introduce a semantic context-free recognition and a language model to acquire similar sequences with high perceptive similarities and semantic correlation, respectively. Moreover, over-confidence degree varies across samples according to their difficulties. Thus, we further design an adaptive calibration intensity module to compute a difficulty score for each samples to obtain finer-grained regularization. Extensive experiments on canonical sequence recognition tasks, including scene text and speech recognition, demonstrate that our method sets novel state-of-the-art results. Code is available at https://github.com/husterpzh/PSSR.

We propose a new method for retrieving 3-D deformation maps by combining the single Interferometric Synthetic Aperture Radar (InSAR)-line of sight (LOS) data and the physical property of crustal deformation, where the spatial coherence of the crustal deformation direction in elastic senses caused by fault slip is considered. First, fault slip and geometry parameters are inverted based on a simplified elastic dislocation model using the InSAR-LOS data. The continuous 3-D surface deformation direction vectors are then estimated by the model parameters. Finally, with the constraints of direction, InSAR-LOS is transformed into the 3-D components, and the 3-D coseismic deformation field is obtained. We implement the L-band SAR data from Advanced Land Observing Satellite (ALOS)/Phased Array L-band SAR (PALSAR) for the 2008 Wenchuan earthquake to testify our method. For the purpose of comparison, four representative models are constructed, ranging from rough to fine. The experimental results explain that the spatial consistency of our method agrees well with the field investigation and Global Navigation Satellite System (GNSS) observations and shows more application potentials than forward modeling.