The reliable operation of Unmanned Aerial Vehicles (UAVs) as mobile Internet of Things (IoT) assets is critically dependent on robust, real-time anomaly detection within their high-dimensional flight data. Conventional methods based on reconstruction error often fail to capture the strong spatial-temporal dependencies and nonlinear dynamics inherent in time-series flight data. To address these limitations, this paper proposes the Nonlinear Latent Dynamic Consistency Probabilistic Variational Auto-Encoder (NLDC-VAE), a novel framework that synergistically integrates a hybrid convolutional-recurrent encoder for spatio-temporal feature extraction with a nonlinear, GRU-based transition network that enforces latent dynamic consistency. Furthermore, a probabilistic decoder enables robust uncertainty quantification by reconstructing a full distribution with mean and variance. A dual anomaly score (DAS) mechanism is introduced, which decouples the assessment of anomalies into two distinct components: the observation anomaly score measures the Negative Log-Likelihood of reconstruction, and the dynamic anomaly score quantifies latent inconsistency via Kullback-Leibler divergence. Experimental results on both simulation and real-world UAV flight data demonstrate that NLDC-VAE achieves superior anomaly detection performance and robustness, confirming its effectiveness for safety-critical UAV monitoring applications.

Large Vision-Language Models (LVLMs) with "multimodal distractibility," where plausible but irrelevant visual or textual inputs cause significant drops in reasoning consistency and lead to unreliable outputs. This paper introduces a comprehensive framework to systematically diagnose, evaluate, and mitigate this critical challenge. We present three core components: the large-scale IR-VQA benchmark to surface these vulnerabilities across four paradigms; novel diagnostic metrics, Positive Consistency (PC) and Negative Consistency (NC), which move beyond standard accuracy to rigorously measure a model's reasoning stability; and the Relevance-Gated Multimodal Routing (RGMR) mechanism, a novel, lightweight module that proactively and dynamically filters distractions at inference time. Our experiments reveal that state-of-the-art models exhibit significant drops in consistency on IR-VQA. We demonstrate that finetuning on IR-VQA and deploying RGMR substantially improve model robustness where standard prompting fails. Our comprehensive analysis of model behaviors under different types of distractions and the underlying reasoning failures provides a clear path forward for developing more reliable multimodal systems.

Clouds in optical remote sensing images (ORSI) significantly limit image utilization. Traditional cloud removal methods using single or multi-temporal data sources struggle to ensure reliable reconstruction for thick cloud areas. Synthetic Aperture Radar (SAR) images are increasingly used to recover information obscured by clouds, but their performance in cloud-obscured regions is unstable. Therefore, an adaptive cloud removal network for remote sensing images, named DG2-TCR, is proposed based on SAR-driven dual-flow fusion guidance (DFG). DG2-TCR uses SAR and ORSI to construct DFG, including local spatial-spectral feature reconstruction (LSSFR) flow and global texture feature compensation (GTFC). LSSFR, driven by ORSI and SAR, efficiently extracts useful features in non-cloud areas and focuses on local information reconstruction using the designed spatial-spectral features inference reconstruction block (SSIRB). Based on SAR images, GTFC guides the compensation of global texture information. DFG can adaptively extract features and reconstruct missing information from local and global scales. The public SEN12MS-CR-TS dataset is divided into four sub-datasets with different coverage to evaluate the recovering capability in varying clouds. Experiments show that the PSNR, SSIM, RMSE, FID, and NCC indicator values on four sub-datasets and the SIMLE-CR dataset are better than the seven comparison methods. Furthermore, the ablation experiments show that the generalization and robustness of this proposed method on images with different cloud coverage are better than other comparison methods. Therefore, DG2-TCR can reliably recover information on cloud occlusions with various coverage and thickness, which is significant for cloud removal in practical applications.

Integrating information from vision and language modalities has sparked interesting applications in the fields of computer vision and natural language processing. Existing methods, though promising in tasks like image captioning and visual question answering, face challenges in understanding real-life issues and offering step-by-step solutions. In particular, they typically limit their scope to solutions with a sequential structure, thus ignoring complex inter-step dependencies. To bridge this gap, we propose a graph-based approach to vision-language problem solving. It leverages a novel integrated attention mechanism that jointly considers the importance of features within each step as well as across multiple steps. Together with a graph neural network method, this attention mechanism can be progressively learned to predict sequential and non-sequential solution graphs depending on the characterization of the problem-solving process. To tightly couple attention with the problem-solving procedure, we further design new learning objectives with attention metrics that quantify this integrated attention, which better aligns visual and language information within steps, and more accurately captures information flow between steps. Experimental results on VisualHow, a comprehensive dataset of varying solution structures, show significant improvements in predicting steps and dependencies, demonstrating the effectiveness of our approach in tackling various vision-language problems.

Cloud detection in high-resolution remote sensing images (HRSI) is widely applied to cross-spatiotemporal domains with various scenarios change. However, cloud detection semantic segmentation models based on limited samples cannot ensure the consistency of data distribution between the source domain (SD) and the target domain (TD), resulting in a decrease in cross-domain segmentation accuracy and robust ability. Therefore, this paper proposed a two-stage domain adaptation based on the image and feature levels (TDAIF) cloud detection framework. TDAIF designs a pseudo-target domain data generator (PTDDG) at the image level to fuse the SD foreground and TD background information effectively, assisting the model in mining invariant semantic knowledge of the TD. Then, a domain discriminator and self-ensembling joint (DDSEJ) framework is explored at the feature level to implicitly handle the alignment of global features and the optimization of decision boundaries-local features. TDAIF ultimately weakens the impact of image radiation diversity and scale divergence and improves the adaptive processing capabilities for cross-spatiotemporal data. Horizontal and internal comparative experiments on TDAIF were conducted on three domain transfer data. Experimental results show that TDAIF dramatically reduces the network accuracy loss in cross-domain. Compared with CycleGAN and AdaptSegNet, the IoU is improved by about 30%. TDAIF performs better than state-of-the-art computational visual domain adaptation methods, indicating that hierarchical data alignment from the image to the feature level is very effective.

As the number of electric vehicles (EVs) increases rapidly, the problem of electric vehicle charging has widely become a concern. Therefore, considering the fact that charging time for one EV cannot be shortened quickly and the number of charging stations will not expand rapidly, how to schedule charging operations of electric vehicles in urban areas becomes a very important issue, since it can improve charging efficiency and relieve charging anxiety of EV users. Up to now, there is no scheduling software tool for practical use in this field. Based on the analysis of electric vehicle charging behavior characteristics, this paper investigates the EV charging problem at the scheduling level. First, a mathematical model for coordinated charging of EVs is proposed to minimize the total charging time for a given number of vehicles. Second, an earliest finish charging scheduling algorithm is presented to solve the charging problem. Then, by considering the combinatorial nature and practical applications with large number of EVs, two practical swarm-optimization-based EV charging scheduling algorithms are proposed. A real-life case study is presented to illustrate the proposed approaches.

Recent progress in the interdisciplinary studies of computer vision (CV) and natural language processing (NLP) has enabled the development of intelligent systems that can describe what they see and answer questions accordingly. However, despite showing usefulness in performing these vision-language tasks, existing methods still struggle in understanding real-life problems (i.e., how to do something) and suggesting step-by-step guidance to solve them. With an overarching goal of developing intelligent systems to assist humans in various daily activities, we propose VisualHow, a free-form and open-ended research that focuses on understanding a real-life problem and deriving its solution by incorporating key components across multiple modalities. We develop a new dataset with 20,028 real-life problems and 102,933 steps that constitute their solutions, where each step consists of both a visual illustration and a textual description that guide the problem solving. To establish better understanding of problems and solutions, we also provide annotations of multimodal attention that localizes important components across modalities and solution graphs that encapsulate different steps in structured representations. These data and annotations enable a family of new vision-language tasks that solve real-life problems. Through extensive experiments with representative models, we demonstrate their effectiveness on training and testing models for the new tasks, and there is significant scope for improvement by learning effective attention mechanisms. Our dataset and models are available at https://github.com/formidify/VisualHow.

While attention has been an increasingly popular component in deep neural networks to both interpret and boost the performance of models, little work has examined how attention progresses to accomplish a task and whether it is reasonable. In this work, we propose an Attention with Reasoning capability (AiR) framework that uses attention to understand and improve the process leading to task outcomes. We first define an evaluation metric based on a sequence of atomic reasoning operations, enabling a quantitative measurement of attention that considers the reasoning process. We then collect human eye-tracking and answer correctness data, and analyze various machine and human attention mechanisms on their reasoning capability and how they impact task performance. To improve the attention and reasoning ability of visual question answering models, we propose to supervise the learning of attention progressively along the reasoning process and to differentiate the correct and incorrect attention patterns. We demonstrate the effectiveness of the proposed framework in analyzing and modeling attention with better reasoning capability and task performance. The code and data are available at https://github.com/szzexpoi/AiR.

While most visual attention studies focus on bottom-up attention with restricted field-of-view, real-life situations are filled with embodied vision tasks. The role of attention is more significant in the latter due to the information overload, and attention to the most important regions is critical to the success of tasks. The effects of visual attention on task performance in this context have also been widely ignored. This research addresses a number of challenges to bridge this research gap, on both the data and model aspects. Specifically, we introduce the first dataset of top-down attention in immersive scenes. The Immersive Question-directed Visual Attention (IQVA) dataset features visual attention and corresponding task performance (i.e., answer correctness). It consists of 975 questions and answers collected from people viewing 360° videos in a head-mounted display. Analyses of the data demonstrate a significant correlation between people's task performance and their eye movements, suggesting the role of attention in task performance. With that, a neural network is developed to encode the differences of correct and incorrect attention and jointly predict the two. The proposed attention model for the first time takes into account answer correctness, whose outputs naturally distinguish important regions from distractions. This study with new data and features may enable new tasks that leverage attention and answer correctness, and inspire new research that reveals the process behind decision making in performing various tasks.

Surveillance video coding is crucial for improving compression efficiency in intelligent video surveillance systems and applications. Plenty of work has been done, which can be roughly divided into two categories: the former mainly focuses on low-complexity background modeling to obtain the clear background, while the latter focuses on an appropriate coding strategy to generate the high-quality background reference picture for effective background prediction. However, almost all existing works focus only on stationary camera scenes, while moving cameras-captured surveillance video coding is left untouched and is still an open problem. In this paper, a background modeling and referencing scheme for moving cameras-captured surveillance video coding in high-efficiency video coding (HEVC) is proposed. First, this paper proposes a low-complexity motion background modeling algorithm for surveillance video coding using the running average based on a global-motion-compensation method. To obtain the global motion vector, we propose a global motion detection method based on character blocks by establishing a low-rank singular value decomposition model for clustering and estimating motion vectors of background character blocks in the cameras movement circumstance. Second, we propose a background referencing coding strategy, in which the motion background coding tree units (MBCTUs) would be selected by anchoring the input video frame on the modeling background frame and coded with the optimized quantization parameter. Then, the reconstructed MBCTU will be used to update the previous coding tree unit in the global compensation location of the background reference picture. Extensive experimental results show that the proposed scheme can achieve significant bit savings of up to 26.6% and, on average, 6.7% with similar subjective quality and negligible encoding complexity, compared to HM12.0. Besides, the proposed scheme consistently outperforms two state-of-the-art surveillance video coding schemes with remarkable bitrate savings.

In this paper, we introduce two improvements on ant colony optimization (ACO) algorithm: route optimization and individual variation. The first is an optimized implementation of ACO, by which the running time of ants routing is largely reduced. The results of the simulated experiments show that the improved algorithm not only reduces the number of routing in the ACO but also surpasses existing algorithms in performance in solving large-scale TSP problems. In the second improvement, we introduce individual variation to ACO, by which the ants have different routing strategies. Simulation results show that the speed of convergence of ACO algorithm could be enhanced greatly.