DCE: Dynamic channel and spatial redundancy elimination for Micro-Expression recognition

Phototropin can trigger the chloroplast accumulation response from multiple subcellular locations in Marchantia polymorpha.

While recent deep learning-based object detection has achieved great success in various fields, it remains challenging to find tiny objects in aerial imagery on-the-fly using mobile devices. Since mobile platforms such as drones operate with limited onboard computing power, handling high-resolution images to find tiny objects with compute-intensive deep learning-based applications often fails to meet their real-time constraints. To mitigate this problem, we propose HashEye, a novel framework that enables fast on-drone tiny object detection by efficiently suppressing spatial redundancy in aerial imagery. HashEye utilizes a lightweight hashing algorithm to rapidly scan image patches; patches exhibiting high hash collision frequencies are identified as background and suppressed. Subsequently, the remaining salient patches are dynamically rearranged into a hardware-friendly dense image for efficient inference. Experimental results on two real-world datasets demonstrate that HashEye achieves up to a 5.25× speedup compared to the baseline, maintaining detection capability.

A modular based framework for multi-step spatio-temporal wind power forecasting

CSA-Kansformer : Cross-scale aggregation and Kansformer network for hyperspectral image classification.

The goal of this study is to empirically evaluate the Decoupled Momentum Optimizer (DeMo) in medical image segmentation while demonstrating its extensibility to applications outside LLMs. We aim to characterize the behavior of each parameter group and their adherence to conjectures underlying DeMo's function. DeMo leverages spatial redundancy in gradients through a spatially partitioned frequency decomposition compression algorithm, reducing network traffic and smoothing gradient noise. DeMo provides up to a 150x traffic reduction and 1.6x wall-time speedup on lung segmentation of COPDGene CTs. Analysis of gradients support the conjectures that the primary components of the gradient exhibited higher spatial autocorrelation and lower temporal variance. We find that these conjectures are not uniformly true across all parameters, but rather are predominantly observed in a small subset of them. We also introduce DeMoDropout, a modification to the algorithm that selectively compresses only the largest gradients to significantly reduce computational overhead while maintaining effective overall compression. Using the Beyond the Cranial Vault dataset, we demonstrate potential speed-ups at bandwidths of 1 Gb/s and 100 Mb/s (1.6x vs 1.5x and 6.151 vs 6.31x for DeMoDropout and DeMo, respectively).

Underwater optical vortex communication faces critical challenges from scattering, turbulence, and transient occlusions, which severely distort orbital angular momentum (OAM) modes. We propose and experimentally demonstrate a robust mode recognition scheme that combines conjugate superimposed OAM beams with a deep residual network (ResNet-50). Nine distinct underwater disturbance environments are quantitatively emulated by independently tuning kaolin concentration, water pump power, and introducing random rectangular occlusions. The petal-like intensity patterns of conjugate superimposed beams preserve discriminative structural information even under strong combined perturbations. Using power-law transformed images as input, ResNet-50 achieves ∼100% classification accuracy for 16 OAM modes across all tested disturbance levels and maintains reliable recognition when occlusions cover up to half of the beam cross-section. The inherent spatial redundancy of conjugate superposition, together with the residual network's feature preservation capability, enables near-perfect generalization without overfitting to experimental artifacts. This work provides a practical, intelligent demodulation framework for deploying highly robust underwater OAM communication systems in real-sea scenarios.

Enhancing SEU tolerance efficacy in advanced FinFET FPGA devices using system-level fine-grained spatial redundancy techniques

Mobility robots for elderly care not only satisfy the basic needs of disabled seniors but also help ensure their safety. Safety monitoring is particularly critical when disabled seniors remain alone indoors. This research focuses on detecting flame and smoke targets in indoor environments, enabling faster decision-making during fires, facilitating timely evacuation for disabled seniors, and thereby providing improved protection. This study aims to enhance detection accuracy and algorithm performance by introducing the improved YOLO11-BSCS model. The Biformer two-layer routing attention mechanism is incorporated into the Backbone and Neck of YOLO11s, replacing the original C2SPA module with C2SPA_Biformer to enable dynamic, query-aware sparse attention, reduce the number of model parameters, and improve the detection of dynamic targets. The SCConv convolution replaces the C3k2 convolution module in the original model with the C3k2_SCConv module, reducing spatial and channel redundancy during the fusion of image features extracted by the model and increasing detection speed. The loss function of the model was optimized by replacing CIoU-Loss with the SIoU-Loss module. This modification improves both convergence speed and detection accuracy. Through 600 rounds of experimental testing on 5,000 data samples, supplemented by three independent training runs using random seeds (107,325,592) for evaluation, YOLO11-BSCS achieved 94.612% accuracy, 89.678% recall, and 90.319% average precision—representing improvements of 4.934, 7.452, and 5.184%, respectively, over YOLO11s. Comparative analysis with widely used models indicates that YOLO11-BSCS provides strong generalizability, precise localization, robust detection, and overall superior performance. The necessity of each model enhancement was validated through ablation experiments, confirming that all modifications contributed meaningfully to performance improvements. These findings provide a valuable reference for addressing similar challenges in object detection.

Light-weight neural networks for remote sensing (RS) visual analysis must overcome two inherent redundancies: spatial redundancy from vast, homogeneous backgrounds, and channel redundancy, where extreme scale variations render a single feature space inefficient. Existing models, often designed for natural images, fail to address this dual challenge in RS scenarios. To bridge this gap, we propose LWGANet, a light-weight backbone engineered for RS-specific properties. LWGANet introduces two core innovations: a Top-K Global Feature Interaction (TGFI) module that mitigates spatial redundancy by focusing computation on salient regions, and a Light-Weight Grouped Attention (LWGA) module that resolves channel redundancy by partitioning channels into specialized, scale-specific pathways. By synergistically resolving these core inefficiencies, LWGANet achieves a superior trade-off between feature representation quality and computational cost. Extensive experiments on twelve diverse datasets across four major RS tasks—scene classification, oriented object detection, semantic segmentation, and change detection—demonstrate that LWGANet consistently outperforms state-of-the-art light-weight backbones in both accuracy and efficiency. Our work establishes a new, robust baseline for efficient visual analysis in RS images.

With the proliferation of Internet of Things (IoT) deployments and mobile sensing systems, reversible data hiding in encrypted images (RDHEI) has emerged as a cornerstone technology for secure cloud-based sensor data management. RDHEI ensures data confidentiality while enabling bit-to-bit restoration of original visual assets. However, conventional RDHEI methods often struggle to optimize the trade-off between high embedding capacity (EC) and the fidelity requirements of sensor-acquired content. This paper proposes an advanced RDHEI framework based on Adaptive Predicted Value Computation and Pixel Classification (APVCPC). The core contribution is a context-aware prediction engine that adaptively selects optimal estimation functions based on local texture complexity, significantly enhancing prediction accuracy in heterogeneous image regions. Subsequently, a content-driven pixel classification paradigm categorizes pixels into loadable (Lpxls) and non-loadable (NLpxls) sets using a dynamic threshold, maximizing the utilization of spatial redundancy. The proposed scheme further supports separable data extraction and image decryption, providing flexible access control for diverse user privileges in secure sensing scenarios. Experimental results on standard benchmarks and the BOW-2 database demonstrate that APVCPC achieves a superior average embedding rate exceeding 2.0 bpp and ensures perfect reversibility, significantly outperforming state-of-the-art techniques in terms of both capacity and security.

<sec><title>Objective</title> Urbanization has fragmented ecological habitats, threatening urban ecosystem sustainability. Ecological networks are crucial for maintaining resilience, with dynamic interactions between network systems and urban development. Amid the green transformation of cities, refining the framework for optimizing urban ecological networks is essential. However, current research mainly focuses on static optimization, neglecting the dynamic evolution between nature and urbanization, and overlooks land use/cover changes in mid-to-small-scale areas, weakening the social functions of ecological networks. This study aims to address the limitations of current static approaches to urban ecological network optimization by investigating the dynamic interplay between urbanization and ecological network resilience. Specifically, focusing on Haizhu District, Guangzhou, this research will discuss the following issues. 1) Simulate future urban development scenarios and potential disturbances, considering spatio-temporal dynamics and land use/cover changes at a mesoscale; 2) Develop and apply a framework for proactively measuring the resilience of urban ecological networks under these dynamic scenarios; 3) Propose targeted design responses and optimization strategies to enhance ecological network resilience, incorporating principles of regional coordination; 4) Ultimately, inform sustainable urban development orientations and operational strategies for Haizhu District, contributing to a more resilient and ecologically sound urban environment. </sec><sec><title>Methods</title> This study develops a theoretical framework to analyze the dynamic coupling between urban development and ecological networks, and proposes a new concept of "Urban-Wetland Complex" and summarizes it along with its characteristics. The framework is applied to Haizhu District, Guangzhou (90.42 km²), a representative wetland urban area in the Pearl River Delta, serving as a case study. Multi-source data, including land use and land cover change (LUCC) data, remote sensing imagery, and socio-economic statistics, are integrated to parameterize the patch-generating land use simulation (PLUS) model. This model is then used to simulate three future development scenarios: 1) natural evolution (baseline scenario without policy intervention), 2) ecological priority (scenario that maximizes ecological benefits), and 3) economic priority (scenario that maximizes economic efficiency). Based on multi-temporal land-use classification maps derived from these simulations, ecological networks are constructed under each scenario. A dual-dimensional resilience assessment model, encompassing both structural and functional aspects, is proposed to quantify the resilience of the network. Structural resilience is evaluated using Graph Theory metrics, specifically connectivity probability and network closure. Functional resilience is assessed through betweenness centrality analysis and node-deletion experiments, focusing on key node identification rate and overall network robustness. </sec><sec><title>Results</title> 1) Structural and functional resilience mechanisms: Structural resilience was primarily influenced by the degree of source fragmentation, the average length of ecological corridors, and network transmissibility. Functional resilience, in contrast, depended on the number of critical nodes and their sensitivity to node removal processes. The comparison of network resilience before and after node removal effectively identified key ecological nodes within the network. 2) Spatial patterns of ecological sources: Ecological sources in Haizhu District were largely consistent with the Pearl River network, wetland parks (e.g., Haizhu Wetland) and existing urban green space systems, reflecting the dominance of wetland−river interactions in shaping the ecological structure. 3) Scenario-based resilience performance: Under the natural development scenario, network connectivity, disturbance resistance, and connection efficiency were the highest, indicating superior structural resilience. The shortened average corridor length enhanced species migration efficiency, leading to improved overall network performance. Under the urban expansion scenario, the ecological network displayed a strong dependence on a limited number of critical nodes, resulting in weakened functional resilience and reduced spatial redundancy. Nodes were highly concentrated south of the wetlands, aligning with the phenomenon of ecological islanding induced by intensive development. Under the ecological priority scenario, the network exhibited the highest resilience threshold and overall stability. Secondary nodes such as Dawei Park and the riverside green belts formed a "core wetland−multi-tiered pivot" structure, enhancing spatial equilibrium and redundancy. However, the longer corridor paths in this scenario were more frequently interrupted by urban infrastructure and human disturbance. </sec><sec><title>Conclusion</title> This study advances on previous research on the identification of key ecological patches and corridors by incorporating a spatiotemporal perspective into the analysis of urban ecological networks. By assessing the evolution of landscape connectivity under multiple development scenarios, this study reveals the dynamic interactions between urban expansion and ecological resilience. The study further refines the methodological framework for evaluating urban ecological networks, establishing a three-dimensional analytical system integrating "network−resilience−potential". This framework not only provides scientific guidance for determining regional ecological development directions and optimizing land-use planning but also serves as a theoretical reference for understanding the dynamic coupling between nature and city in high-density urban environments. Ultimately, the findings contribute to bridging the gap between ecological modeling and spatial design, supporting the construction of adaptive, resilient, and sustainable urban ecological systems. </sec>

Down-sampling-based video compression frameworks have shown great potential in improving compression efficiency in modern sensing and imaging systems. However, existing methods ignore critical spatial and temporal redundancy, and treat all frames uniformly during down-sampling. This leads to the loss of important information and impacts compression efficiency. To address these limitations, this paper proposes a temporal down-sampling system, in which only intermediate frames are down-sampled while preserving key frames with high quality for reference. On the decoding side, we employ a frame-recurrent enhancement mechanism to maximize the use of temporal redundancy information. In the fusion of enhancement stage, we design a Multi-scale Temporal-Spatial Attention (MTSA) module. MTSA consists of two components: Multi-Temporal Attention (MTA) and Pyramid Spatial Attention (PSA). MTA performs multi-scale temporal correlation modeling, expanding the receptive field and providing stable cues in compressed regions. PSA integrates local spatial saliency and contextual structure in a progressive and multi-stage manner. Extensive experiments show that our approach achieves consistent BD-rate reductions. Under All-Intra, Low-Delay-P, and Random Access configurations, we observe BD-rate reductions of I, P, and B frames ranging from 14% to 39% compared to VVC, and outperform prior approaches anchored by the standard HEVC.

Magnetic resonance imaging (MRI) benefits significantly from parallel imaging and low-rank matrix completion approaches to reconstruct accelerated multidimensional acquisitions. As one such algorithm, low-rank matrix modelling of local k-space neighbourhoods (LORAKS) provides image reconstruction by exploiting sparsity in undersampled multichannel data acquisitions. Besides using spatial and multichannel redundancies, joint-reconstruction of multi-contrast data provides higher-quality reconstruction and enables higher acceleration factors. However, the computational demands of LORAKS limit its applicability for joint-reconstruction of modern multichannel multi-contrast high-resolution datasets, but methods to speed up the acquisition and reconstruction process are highly desired. In this work, we present PyLORAKS, a graphics processing unit (GPU) accelerated implementation of LORAKS built using the PyTorch framework. The framework employs efficient parallelisation, GPU compute strategies, and randomised singular value decomposition methods for increased computational efficiency in dealing with huge LORAKS matrix sizes encountered in modern high-resolution MRI. The computation speedup enabled parameter optimisation via Bayesian methods, while automated differentiable tracing of the LORAKS formulation was used for data-driven subsampling optimisation. In benchmark experiments, reduction in memory overhead was achieved, paired with approximately 5–6× speedup over the previous (MATLAB) LORAKS implementations using CPU computation, and a ~30-fold improvement using PyLORAKS on GPUs. While ensuring similar reconstruction performance, the computational efficiency allows for reconstruction of larger data matrices previously posing challenges to memory or computation time demands. Automated parameter optimisation of LORAKS parameters λ and r yielded better results than manual selection, avoiding inter-contrast leakage artefacts. Additionally, the highest reconstruction quality was achieved by using the maximal number of available contrasts for joint-reconstruction, exemplarily resulting in an increased structural similarity (SSIM) from 0.9468 to 0.9688 when doubling the number of echoes in joint-reconstruction of a 4-fold undersampled multi-echo acquisition. Furthermore, complementary sampling across echoes was found beneficial as reconstruction of the same sampling per echo achieved an SSIM of 0.955, whereas interleaving phase encode lines per echo increased the SSIM to 0.972 for the same acquisition and otherwise optimal parameter combinations. We further demonstrate superior performance of complementary sampling using an optimal sampling pattern obtained by backpropagation through the PyLORAKS algorithm. Overall, PyLORAKS enables fast and efficient reconstruction of multichannel multi-contrast high-resolution MRI data. It enables parameter and data-driven reconstruction and acquisition optimisation and provides a platform for physics-informed or artificial intelligence-augmented development. The framework is openly available, including containerised environments for large-scale deployment.

Understanding the distribution and conservation status of small carnivores is critical for informing management strategies in human-modified landscapes. We assembled a comprehensive dataset of jungle cat (Felis chaus) presence across India, drawing from over 26,000 camera trap locations, radio-telemetry data, published literature, secondary sources, and verified sightings. After filtering for spatial redundancy, we modeled species distribution using ecologically relevant covariates in both maximum entropy (MaxEnt) and random forest (RF) frameworks. The resulting ensemble model indicated that jungle cats are most likely to occur in warm, semi-arid regions with moderate vegetation cover and low to moderate levels of human and livestock disturbance. In contrast, they tend to avoid dense forests and highly transformed habitats. Despite their broad geographic distribution, jungle cats face increasing threats from habitat fragmentation, expanding infrastructure, road mortality, disease transmission from free-ranging dogs, and genetic introgression through hybridization with domestic cats. These pressures are particularly acute in peri-urban and agro-pastoral landscapes where jungle cats persist outside protected areas. Our findings underscore the importance of rural lifestyles with agro-pastoralism livelihoods for conserving the species along with grasslands, savanna and open forest systems to ensure the species’ long-term viability in a rapidly urbanizing landscape.

Cyclone Global Navigation Satellite System (CYGNSS) has emerged as an effective technique for inland water body detection due to its high sensitivity to inland waters. However, existing methods for inland water body detection using CYGNSS are limited by the difficulty in balancing high spatiotemporal resolution with strong generalization capability. Moreover, the limited spatial redundancy in short-term CYGNSS data restricts its capacity for high-precision inland water detection on its own. To address these issues, this study proposed a novel dual-branch model, termed STRUE. The model integrated a Swin Transformer and ResNet within a U-Net-enhanced student-teacher framework. This framework was developed through the fusion of multi-source data, including CYGNSS, SMAP, FABDEM, MODIS, and GSWE. The results showed that, for inland water body detection, the model attained a spatial resolution of 0.01° and a temporal resolution of 7 days. In terms of performance, it achieved an F1-score (F1) of 0.914, a mean Intersection over Union (mIoU) of 0.880, a Matthews Correlation Coefficient (MCC) of 0.873, and a Recall (R) of 0.963. Additionally, compared with traditional methods and models, the proposed model demonstrated a better performance in spatial continuity, structural integrity, and detail recovery, while mitigating common limitations such as cloud obscuration, spatial incoherence, and overestimation artifacts. These results further enhance the capacity of spaceborne GNSS-R for inland water body detection.

The impact of spatial redundancy and merging conflicts on crowd movement patterns: An empirical study

A well-structured urban park system (UPS) is crucial for optimizing urban spatial layout and improving the quality of the human living environment. In response to the tendency of current planning to prioritize quantitative indicators while overlooking the relational structure arising from the collective spatial configuration of parks, this study introduces Social Network Analysis (SNA) to evaluate the spatial structure of Shanghai’s park system by constructing a service-coverage overlap network. The findings reveal the following: (1) Parks with high degree centrality are concentrated in high-density urban core areas due to service overlap, whereas large suburban parks with high betweenness centrality function as critical bridging hubs, reflecting a polycentric structure. (2) There is a discernible discrepancy between these emergent network tiers and the statutory park hierarchy, highlighting a tension between bottom-up spatial patterns and top-down planning frameworks. (3) Stability simulations indicate a dual character of the system, where the network topology is vulnerable to attacks yet functionally resilient to failures due to spatial redundancy, suggesting that a decline in service quality may precede the loss of basic accessibility. This study demonstrates the value of SNA in diagnosing park system structure, identifying key nodes, and assessing system resilience. The insights advocate for planning approaches that transcend rigid hierarchical frameworks, integrate the actual functional roles of parks, and protect structural hubs, thereby enhancing systemic resilience and promoting equitable service provision.

This paper presents a lightweight hybrid framework that integrates a Haar-initialized Parametric Wavelet Transform (PWT) with a Convolutional Neural Network (CNN) enhanced by a multi-head Self-Attention mechanism for efficient and interpretable tumor identification from compressed Magnetic Resonance Imaging (MRI) brain image data. A Parametric Wavelet Transform (PWT) layer, initialized with Haar wavelet filters, performs compression and adaptive feature extraction from brain MRI images, enabling the model to learn optimal frequency decompositions while preserving diagnostic features. MRI images are preprocessed through this PWT layer to selectively extract and stack the approximation and diagonal detail subbands, reducing spatial redundancy and enhancing the representation of diagnostically salient structures. A custom lightweight CNN backbone extracts local features from frequency-domain representations. The integrated self-attention module captures salient features and enhances the discriminative power across wavelet-transformed inputs. Grad-CAM visualizations focussed on explaining the model's predictions and attended to tumor relevant regions. The primary contribution of the proposed model focuses on the overall performance with a classification accuracy of 95.88%, which is higher than the benchmark models of MobileNetV2 (93.1%) and MobileNetV3Small (94.80%) while preserving less trainable parameters and memory footprint. An ablation study confirms the individual contributions towards the overall model performance of PWT compression, the CNN backbone, and the self-attention module. Deploying the model on a Raspberry Pi 5 highlights the potential for real-time, point-of-care, edge-based medical imaging. This work is a pioneering integrated approach incorporating adaptive frequency-domain compression alongside attention-based refinement to produce interpretable and robust designs for embedded implementations of brain tumor classification.

Linear array detector-based infrared push-broom imaging systems are widely employed in remote sensing and security surveillance due to their high spatial resolution, wide swath coverage, and low cost. However, the massive data volume generated during continuous scanning presents substantial storage and transmission challenges. To mitigate this issue, we propose a lossless compression algorithm based on pixel-adaptive prediction and hierarchical decomposition of residuals. The algorithm first performs pixel-wise adaptive noise compensation according to local image characteristics and achieves efficient prediction by exploiting the strong inter-pixel correlation along the scanning direction. Subsequently, hierarchical decomposition is applied to high-energy residual blocks to further eliminate spatial redundancy. Finally, the Golomb–Rice coding parameters are adaptively adjusted based on the neighborhood residual energy, optimizing the overall code length distribution. The experimental results demonstrate that our method significantly outperforms most state-of-the-art approaches in terms of both the compression ratio (CR) and bits per pixel (BPP). Moreover, while maintaining a CR comparable to H.265-Intra, our method achieves a 21-fold reduction in time complexity, confirming its superiority for large-format image compression.