Current dark image restoration methods suffer from severe efficiency bottlenecks, primarily stemming from: computational burden and error correction costs associated with reliance on external priors (manual or cross-modal); redundant operations in complex multi-stage enhancement pipelines; and indiscriminate processing across frequency components in frequency-domain methods, leading to excessive global computational demands. To address these challenges, we propose an Efficient Self-Mining Prior-Guided Joint Frequency Enhancement Network (SPJFNet). Specifically, we first introduce a Self-Mining Guidance Module (SMGM) that generates lightweight endogenous guidance directly from the network, eliminating dependence on external priors and thereby bypassing error correction overhead while improving inference speed. Second, through meticulous analysis of different frequency domain characteristics, we reconstruct and compress multi-level operation chains into a single efficient operation via lossless wavelet decomposition and joint Fourier-based advantageous frequency enhancement, significantly reducing parameters. Building upon this foundation, we propose a Dual-Frequency Guidance Framework (DFGF) that strategically deploys specialized high/low frequency branches (wavelet-domain high-frequency enhancement and Fourier-domain low-frequency restoration), decoupling frequency processing to substantially reduce computational complexity. Rigorous evaluation across multiple benchmarks demonstrates that SPJFNet not only surpasses state-of-the-art performance but also achieves significant efficiency improvements, substantially reducing model complexity and computational overhead.

Recovering fine-grained details in extremely dark images remains challenging due to severe structural information loss and noise corruption. Existing enhancement methods often fail to preserve intricate details and sharp edges, limiting their effectiveness in downstream applications like text and edge detection. To address these deficiencies, we propose an efficient dual-stage approach centered on detail recovery for dark images. In the first stage, we introduce a Residual Fourier-Guided Module (RFGM) that effectively restores global illumination in the frequency domain. RFGM captures inter-stage and inter-channel dependencies through residual connections, providing robust priors for high-fidelity frequency processing while mitigating error accumulation risks from unreliable priors. The second stage employs complementary Mamba modules specifically designed for textural structure refinement: (1) Patch Mamba operates on channel-concatenated non-downsampled patches, meticulously modeling pixel-level correlations to enhance fine-grained details without resolution loss. (2) Grad Mamba explicitly focuses on high-gradient regions, alleviating state decay in state space models and prioritizing reconstruction of sharp edges and boundaries. Extensive experiments on multiple benchmark datasets and downstream applications demonstrate that our method significantly improves detail recovery performance while maintaining efficiency. Crucially, the proposed modules are lightweight and can be seamlessly integrated into existing Fourier-based frameworks with minimal computational overhead.

Urban region profiling, the task of characterizing geographical areas, is crucial for urban planning and resource allocation. However, existing research in this domain faces two significant limitations. First, most methods are confined to single-task prediction, failing to capture the interconnected, multi-faceted nature of urban environments where numerous indicators are deeply correlated. Second, the field lacks a standardized experimental benchmark, which severely impedes fair comparison and reproducible progress. To address these challenges, we first establish a comprehensive benchmark for multi-task urban region profiling, featuring multi-modal features and a diverse set of strong baselines to ensure a fair and rigorous evaluation environment. Concurrently, we propose UrbanMoE, the first sparse multi-modal, multi-expert framework specifically architected to solve the multi-task challenge. Leveraging a sparse Mixture-of-Experts architecture, it dynamically routes multi-modal features to specialized sub-networks, enabling the simultaneous prediction of diverse urban indicators. We conduct extensive experiments on three real-world datasets within our benchmark, where UrbanMoE consistently demonstrates superior performance over all baselines. Further in-depth analysis validates the efficacy and efficiency of our approach, setting a new state-of-the-art and providing the community with a valuable tool for future research in urban analytics.

Traditional Low-Light Image Enhancement (LLIE) methods primarily focus on uniform brightness adjustment, often neglecting instance-level semantic information and the inherent characteristics of different features. To address these limitations, we propose CWNet (Causal Wavelet Network), a novel architecture that leverages wavelet transforms for causal reasoning. Specifically, our approach comprises two key components: 1) Inspired by the concept of intervention in causality, we adopt a causal reasoning perspective to reveal the underlying causal relationships in low-light enhancement. From a global perspective, we employ a metric learning strategy to ensure causal embeddings adhere to causal principles, separating them from non-causal confounding factors while focusing on the invariance of causal factors. At the local level, we introduce an instance-level CLIP semantic loss to precisely maintain causal factor consistency. 2) Based on our causal analysis, we present a wavelet transform-based backbone network that effectively optimizes the recovery of frequency information, ensuring precise enhancement tailored to the specific attributes of wavelet transforms. Extensive experiments demonstrate that CWNet significantly outperforms current state-of-the-art methods across multiple datasets, showcasing its robust performance across diverse scenes. Code is available at https://github.com/bywlzts/CWNet-Causal-Wavelet-Network.

Low-light image enhancement presents two primary challenges: 1) Significant variations in low-light images across different conditions, and 2) Enhancement levels influenced by subjective preferences and user intent. To address these issues, we propose ReF-LLE, a novel personalized low-light image enhancement method that operates in the Fourier frequency domain and incorporates deep reinforcement learning. ReF-LLE is the first to integrate deep reinforcement learning into this domain. During training, a zero-reference image evaluation strategy is introduced to score enhanced images, providing reward signals that guide the model to handle varying degrees of low-light conditions effectively. In the inference phase, ReF-LLE employs a personalized adaptive iterative strategy, guided by the zero-frequency component in the Fourier domain, which represents the overall illumination level. This strategy enables the model to adaptively adjust low-light images to align with the illumination distribution of a user-provided reference image, ensuring personalized enhancement results. Extensive experiments on benchmark datasets demonstrate that ReF-LLE outperforms state-of-the-art methods, achieving superior perceptual quality and adaptability in personalized low-light image enhancement.

Goal-conditioned reinforcement learning (GCRL) is an effective method for multi-goal robotic manipulation tasks. Many studies based on hindsight experience replay (HER) and hindsight goal generation (HGG) have achieved the autonomous acquisition of robotic manipulation in reward-sparse environments and have greatly improved the learning efficiency of GCRL. However, these methods perform poorly in environments with obstacles and distant goals. In this paper, we propose hindsight goal diffusion and graph-based experience replay (HGD-GER) for complex robotic manipulation. First, obstacle-avoiding graphs in environments with obstacles are constructed, and the graph-based distance metric between different goals is established. Second, the proposed HGD approach utilizes the inherent denoising mechanism of diffusion models and obstacle-avoiding graph-based distance to generate exploration goals, thereby promoting the exploration of obstacle-bypassing areas. Then, GER module modifies the reward value of experience replay by graph-based distance, thereby avoiding the bias introduced by HER and improving the learning performance of the RL algorithm under sparse reward conditions. Finally, we conducted experiments on three robotic manipulation tasks with obstacles and distant goals, and the results show that the proposed HGD-GER achieves excellent learning performance. Additionally, the proposed method is deployed on the physical robot.

In recent years, infrared small target detection (ISTD) networks based on deep learning have achieved notable advances. However, these methods still face significant challenges when applied to the real world. Most of them lack the fundamental principles of small target detection in infrared imagery, which results in difficulties in distinguishing targets from complex backgrounds and poor interpretability. To address these challenges, an interpretable network architecture for ISTD, termed low-rank sparse decomposition state space models (LSDSSMs), is proposed. LSDSSMs use the principles of low-rank and sparse decomposition, incorporating dedicated modules for the low-rank space separation module and the sparse target extraction module. These modules facilitate the extraction of sparse representations for both low-rank backgrounds and small targets. In addition, a joint reconstruction module is employed to integrate these components, generating reconstructed images. Considering the unique imaging characteristics of infrared images and the sparse nature of small targets, a channel selection module (CSM) is proposed to enhance the extraction of sparse targets. To enhance the adaptability, stability, and resistance resistance of LSDSSMs in complex environments, robust state space models are integrated that combine local and global information representations. Furthermore, a multilevel loss function is introduced to enforce comprehensive constraints on low-rank backgrounds, sparse targets, and reconstructed images. This design improves not only the robustness of the LSDSSMs but also its performance across different scenarios. Extensive experimental results demonstrate that LSDSSMs surpass existing baseline methods in both qualitative and quantitative assessments, validating their effectiveness and reliability.

In the Fourier frequency domain, luminance information is primarily encoded in the amplitude component, while spatial structure information is significantly contained within the phase component. Existing low-light image enhancement techniques using Fourier transform have mainly focused on amplifying the amplitude component and simply replicating the phase component, an approach that often leads to color distortions and noise issues. In this paper, we propose a Dual-Stage Multi-Branch Fourier Low-Light Image Enhancement (DMFourLLIE) framework to address these limitations by emphasizing the phase component's role in preserving image structure and detail. The first stage integrates structural information from infrared images to enhance the phase component and employs a luminance-attention mechanism in the luminance-chrominance color space to precisely control amplitude enhancement. The second stage combines multi-scale and Fourier convolutional branches for robust image reconstruction, effectively recovering spatial structures and textures. This dual-branch joint optimization process ensures that complex image information is retained, overcoming the limitations of previous methods that neglected the interplay between amplitude and phase. Extensive experiments across multiple datasets demonstrate that DMFourLLIE outperforms current state-of-the-art methods in low-light image enhancement.

The advancements in deep learning methods have presented new opportunities and challenges for predicting land surface variables (LSVs) due to their similarity with computer sciences tasks. However, few researchers focus on the benchmark datasets for LSVs predictions that hampers fair comparisons of different data-driven deep learning models. Hence, we propose a LSVs benchmark dataset and prediction toolbox to boost research in data-driven LSVs modeling and improve the consistency of data-driven deep learning models for LSVs. LSVs benchmark dataset contains a large number of hydrology-related variables, such as global soil moisture, runoff, etc., which can verify the simulation of hydrological processes. Various global data from European Centre for Medium-Range Weather Forecasts reanalysis 5 (ERA5), ERA5-land, global gridded soil information (SoilGrid), soil moisture storage capacity (SMSC), and moderate-resolution imaging spectroradiometer (MODIS) datasets have been pre-processed into daily data at 0.5-, 1-, 2-, and 4-degree resolutions to facilitate their use in data-driven models. Simple statistical metrics, i.e., the root mean squared error and correlation coefficient, are chosen to evaluate the performance of different deep learning (DL) models, including convolutional neural network, long short-term memory and convolution long short-term memory models, with lead times of 1 and 5 days. A processed-based model serves as a physic baseline, soil moisture and surface sensible heat fluxes are taken as the target variables. The developed benchmark dataset and evaluation metrics for predicting LSVs using data-driven approaches, named as the LandBench toolbox, were implemented using Pytorch. This toolbox facilitates the reimplementation of existing methods, the development of novel predictive models, and the utilization of unified evaluation metrics. Additionally, the toolbox incorporates address mapping technology to enable high-resolution global predictions with constrained computing resources. We hope LandBench will not only serves as a standardized framework, fostering equitable model comparisons, but also provides indispensable data and a robust scientific foundation essential for advancing climate change research, disaster management, and sustainable development initiatives.

Abstract Attention mechanisms are popular techniques in computer vision that mimic the ability of the human visual system to analyse complex scenes, enhancing the performance of convolutional neural networks (CNN). In this paper, the authors propose a refined global attention module (RGAM) to address known shortcomings of existing attention mechanisms: (1) Traditional channel attention mechanisms are not refined enough when concentrating features, which may lead to overlooking important information. (2) The 1‐dimensional attention map generated by traditional spatial attention mechanisms make it difficult to accurately summarise the weights of all channels in the original feature map at the same position. The RGAM is composed of two parts: refined channel attention and refined spatial attention. In the channel attention part, the authors used multiple weight‐shared dilated convolutions with varying dilation rates to perceive features with different receptive fields at the feature compression stage. The authors also combined dilated convolutions with depth‐wise convolution to reduce the number of parameters. In the spatial attention part, the authors grouped the feature maps and calculated the attention for each group independently, allowing for a more accurate assessment of each spatial position’s importance. Specifically, the authors calculated the attention weights separately for the width and height directions, similar to SENet, to obtain more refined attention weights. To validate the effectiveness and generality of the proposed method, the authors conducted extensive experiments on four distinct medical image segmentation datasets. The results demonstrate the effectiveness of RGAM in achieving state‐of‐the‐art performance compared to existing methods.

In this paper, a nonlinear time series analysis method is used to study and analyses crude oil price forecasting, from the perspective of multi-scale analysis, the index is used to construct proxy variables of interest to investors, and the Ensemble Empirical Mode Decomposition (EEMD) method is applied to decompose the crude oil price and Baidu index time series into several independent and different scales of the superposition of modular functions and residual terms, respectively, to extract the volatility characteristics of the series at different time scales. The underlying modal components are divided into high-frequency components, low-frequency components, and residual terms according to the frequency, representing short-term fluctuations, medium-term trends, and long-term trends, respectively. Then, a multi-scale Granger causality test model based on EEMD and a multi-scale sign transfer first method are constructed. Finally, the multi-scale causal relationship between crude oil prices and investors’ attentions in different market capitalizations is systematically investigated by dividing crude oil by market capitalization. Firstly, EEMD decomposes the original time series of oil price and uses the wavelet threshold denoising method to obtain the effective information of high-frequency modal components; secondly, the decomposed modal components are reconstructed using the fine-to-coarse method to obtain the reconstructed components from high to low; then the reconstructed components are predicted using the neural network; finally, the results are obtained by simple summation of the reconstructed series. Compared with the LSTM integrated prediction model with EEMD decomposition, the proposed model improves the oil price prediction accuracy. The two decomposition-integrated deep learning oil price forecasting models proposed in this paper use degree crude oil prices for empirical analysis, and both improve the accuracy of oil price forecasting to some extent. The empirical results show the effectiveness of these two forecasting methods in nonlinear irregular time series forecasting.

Abstract Intracranial haemorrhage (ICH) is a haemorrhagic disease that occurs in the ventricle or brain tissue and has a high probability of mortality and disability. For ICH, it is important to obtain a correct diagnosis in the early stages. Currently, ICH classification mainly depends on professional radiologists for manual diagnosis. Therefore, it is necessary to develop a method that can efficiently and rapidly diagnose ICH. In the field of ICH subtype classification, most studies directly use the existing convolutional neural network (CNN) to extract CT slice features. However, these existing networks have the following shortcomings: (1) insufficient discrimination of CT slice features leads to an inability to achieve satisfactory classification performance. (2) Most CT slice data sets of ICH have the serious problem of sample imbalance. (3) There is a correlation between subtypes; however, in previous studies, this correlation has been ignored. To solve these problems, the authors propose a classification algorithm for ICH subtypes applied to CT images. The CNN–RNN architecture was adopted to classify ICH subtypes. In the CNN module, the problem is viewed from a fine‐grained perspective, which solves the problem of insufficient feature discrimination in existing methods. A new loss function is also proposed to solve the problems of unbalanced data distribution and neglected dependencies among the labels. These parts are integrated into the proposed fine‐grained network architecture. The image embeddings were obtained by the CNN module and then input to the RNN module. The authors’ method was evaluated on the Radiological Society of North America 2019 Brain CT Haemorrhage (RSNA‐2019) benchmark. The experimental results demonstrated that the performance of the proposed method is state‐of‐the‐art.

Moving ground-target detection system is widely used to monitor illegal activities of pedestrians and vehicles. However, existing detection methods are restricted by the power consumption in hardware and are usually based on some single feature of the seismic signal, which leads to low detection accuracy and false alarms. To address these issues, we propose a new moving ground-target detection method for detecting the weak seismic signals generated by distant moving ground targets. This method combines an adaptive strategy and support vector machines (SVMs). Both time- and frequency-domain features of seismic signals are considered in the detection method. Additionally, we carry out field experiments to evaluate the performance of the proposed method. The results show that the proposed moving ground-target detection method can detect distant moving ground targets and avoid false alarms as many as possible, which indicates good performance.

In contrast to the high sensitivity of human eyes and rapid development of modern display devices in terms of dynamic range, mainstream multimedia sources are generally at relatively lower bit depths (BDs). Therefore, BD enhancement (BDE), which attempts to transform low-BD multimedia sources into high-BD sources, is considered of significant research value. Current BDE algorithms are based on images rather than videos. However, for massive numbers of videos, temporal continuity among frames should be considered. Thus, in this paper, we propose a spatiotemporal symmetric BDE network for videos based on an encoder-decoder network. Consecutive frames are input into five subnets in the encoder, where the convolutional filters in the temporal symmetric subnets share the same weights to achieve lower model complexity. In addition, symmetric skip connections are introduced between the symmetric convolutional/deconvolutional layers of the encoder/decoder to pass features and alleviate the gradient diffusion problem. The experimental results show that our model can efficiently eliminate false contours and chroma distortions. The model significantly outperforms state-of-the-art image BDE algorithms and single-frame baseline models in terms of PSNR and SSIM.