The sparse cross-modality detector offers more advantages than its counterpart, the Bird’s-Eye-View (BEV) detector, particularly in terms of adaptability for downstream tasks and computational cost savings. However, existing sparse detectors overlook the quality of token representation, leaving it with a sub-optimal foreground quality and limited performance. In this paper, we identify that the geometric structure preserved and the class distribution are the key to improving the performance of the sparse detector, and propose a Sparse Selector (SS). The core module of SS is Ray-Aware Supervision (RAS), which preserves rich geometric information during the training stage, and Class-Balanced Supervision, which adaptively reweights the salience of class semantics, ensuring that tokens associated with small objects are retained during token sampling. Thereby, outperforming other sparse multi-modal detectors in the representation of tokens. Additionally, we design Ray Positional Encoding (Ray PE) to address the distribution differences between the LiDAR modality and the image. Finally, we integrate the aforementioned module into an end-to-end sparse multi-modality detector, dubbed CrossRay3D. Experiments show that, on the challenging nuScenes benchmark, CrossRay3D achieves state-of-the-art performance with 72.4% mAP and 74.7% NDS, while running$1.84\times $faster than other leading methods. Moreover, CrossRay3D demonstrates strong robustness even in scenarios where LiDAR or camera data are partially or entirely missing. The code is available onhttps://github.com/xuehaipiaoxiang/CrossRay3D

Advanced driver assistance systems require a comprehensive understanding of the driver’s mental/physical state and traffic context but existing works often neglect the potential benefits of joint learning between these tasks. This paper proposes MMTL-UniAD, a unified multi-modal multitask learning framework that simultaneously recognizes driver behavior (e.g., looking around, talking), driver emotion (e.g., anxiety, happiness), vehicle behavior (e.g., parking, turning), and traffic context (e.g., traffic jam, traffic smooth). A key challenge is avoiding negative transfer between tasks, which can impair learning performance. To address this, we introduce two key components into the framework: one is the multi-axis region attention network to extract global context-sensitive features, and the other is the dual-branch multimodal embedding to learn multi-modal embeddings from both task-shared and task-specific features. The former uses a multi-attention mechanism to extract task-relevant features, mitigating negative transfer caused by task-unrelated features. The latter employs a dual-branch structure to adaptively adjust task-shared and task-specific parameters, enhancing cross-task knowledge transfer while reducing task conflicts. We assess MMTL-UniAD on the AIDE dataset, using a series of ablation studies, and show that it outperforms state-of-the-art methods across all four tasks. The code is available on https://github.com/Wenzhuo-Liu/MMTL-UniAD.

Multimodal pre-trained models like CLIP need large image-text pairs for training but often struggle with domain-specific tasks. Since retraining with specialized and historical data incurs significant memory and time costs, it is important to continually learn new domains in the open world while preserving original performance. However, current continual learning research mainly focuses on single-modal scenarios, and the evaluation criteria are insufficient without considering image-text matching performance and the forgetting of zero-shot performance. This work introduces image-caption datasets from various domains and establishes a multimodal vision-language continual learning benchmark. Then, a novel framework named C-CLIP is proposed, which not only prevents forgetting but also enhances new task learning impressively. Comprehensive experiments demonstrate that our method has strong continual learning ability across different domain image-text datasets, and has little forgetting of the original capabilities of zero-shot prediction, significantly outperforming existing methods.

Multi-task learning (MTL) can advance assistive driving by exploring inter-task correlations through shared representations. However, existing methods face two critical limitations: single-modality constraints limiting comprehensive scene understanding and inefficient architectures impeding real-time deployment. This paper proposes TEM3-Learning (Time-Efficient Multimodal Multi-task Learning), a novel framework that jointly optimizes driver emotion recognition, driver behavior recognition, traffic context recognition, and vehicle behavior recognition through a two-stage architecture. The first component, the mamba-based multi-view temporal-spatial feature extraction subnetwork (MTS-Mamba), introduces a forward-backward temporal scanning mechanism and global-local spatial attention to efficiently extract low-cost temporal-spatial features from multi-view sequential images. The second component, the MTL-based gated multimodal feature integrator (MGMI), employs task-specific multi-gating modules to adaptively highlight the most relevant modality features for each task, effectively alleviating the negative transfer problem in MTL. Evaluation on the AIDE dataset, our proposed model achieves state-of-the-art accuracy across all four tasks, maintaining a lightweight architecture with fewer than 6 million parameters and delivering an impressive 142.32 FPS inference speed. Rigorous ablation studies further validate the effectiveness of the proposed framework and the independent contributions of each module. The code is available on https://github.com/Wenzhuo-Liu/TEM3-Learning.

Embodied agents are expected to perform object navigation in dynamic, open-world environments. However, existing approaches typically rely on static trajectories and a fixed set of object categories during training, overlooking the real-world requirement for continual adaptation to evolving scenarios. To facilitate related studies, we introduce the continual object navigation benchmark, which requires agents to acquire navigation skills for new object categories while avoiding catastrophic forgetting of previously learned knowledge. To tackle this challenge, we propose C-Nav, a continual visual navigation framework that integrates two key innovations: (1) A dual-path anti-forgetting mechanism, which comprises feature distillation that aligns multi-modal inputs into a consistent representation space to ensure representation consistency, and feature replay that retains temporal features within the action decoder to ensure policy consistency. (2) An adaptive sampling strategy that selects diverse and informative experiences, thereby reducing redundancy and minimizing memory overhead. Extensive experiments across multiple model architectures demonstrate that C-Nav consistently outperforms existing approaches, achieving superior performance even compared to baselines with full trajectory retention, while significantly lowering memory requirements. The code will be publicly available at \url{https://bigtree765.github.io/C-Nav-project}.

During the process of driving, humans usually rely on multiple senses to gather information and make decisions. Analogously, in order to achieve embodied intelligence in autonomous driving, it is essential to integrate multidimensional sensory information in order to facilitate interaction with the environment. However, the current multi-modal fusion sensing schemes often neglect these additional sensory inputs, hindering the realization of fully autonomous driving. This paper considers multi-sensory information and proposes a multi-modal interactive perception dataset named MIPD, enabling expanding the current autonomous driving algorithm framework, for supporting the research on embodied intelligent driving. In addition to the conventional camera, lidar, and 4D radar data, our dataset incorporates multiple sensor inputs including sound, light intensity, vibration intensity and vehicle speed to enrich the dataset comprehensiveness. Comprising 126 consecutive sequences, many exceeding twenty seconds, MIPD features over 8,500 meticulously synchronized and annotated frames. Moreover, it encompasses many challenging scenarios, covering various road and lighting conditions. The dataset has undergone thorough experimental validation, producing valuable insights for the exploration of next-generation autonomous driving frameworks. Data, development kit and more details will be available athttps://github.com/BUCT-IUSRC/Dataset__MIPD

In recent years, researchers have focused on identifying tasks related to driver state, traffic environment, and others to enhance the safety of autonomous driving assistance systems. However, current research on these tasks is conducted independently, neglecting the interconnections between the driver, traffic environment, and vehicle. In this paper, we propose a Unified Multi-task Assistive Driving Network Based on Multimodal Fusion (UMD-Net), the first unified model capable of recognizing four tasks simultaneously by utilizing multimodal data: driver behavior recognition, driver emotion recognition, traffic context recognition, and vehicle behavior recognition. In order to better enhance the synergistic effects between multiple tasks, we designed the position-sensitive multi-directional attention feature extraction subnetwork and recursive dynamic feature fusion module. The former captures the key features of multi-view images by different directions of attention mechanism to improve the generalization of the model across multiple tasks. The latter dynamically adjusts the fusion weight according to the multimodal features to enhance the representation ability of important features in multi-task learning. Our model was evaluated on the public dataset AIDE, achieving the best performance across all four tasks and a high accuracy of 95.31% in the traffic context recognition task, demonstrating the superiority of our approach. The code is available on https://github.com/Wenzhuo-Liu/UMD-Net.

The self-supervised learning (SSL) has emerged as an effective paradigm for deriving general representations from vast amounts of unlabeled data. However, as real-world applications continually integrate new content, the high computational and resource demands of SSL necessitate continual learning (CL) rather than complete retraining. This poses a challenge in balancing between stability and plasticity when adapting to new information. In this article, we employ centered kernel alignment (CKA) for quantitatively analyzing model stability and plasticity, revealing the critical roles of batch normalization (BN) layers for stability and convolutional layers for plasticity. Motivated by this, we propose branch-tuning (BT), an efficient and straightforward method that achieves a balance between stability and plasticity in continual SSL. BT consists of branch expansion and compression and can be easily applied to various SSL methods without the need of modifying the original methods, retaining old data or models. We validate our method through experiments on various benchmark datasets, demonstrating its effectiveness and practical value in real-world scenarios. We hope our work offers new insights for future continual SSL research. The code will be made publicly available.

Acquiring large-scale, well-annotated datasets is essential for training robust scene text detectors, yet the process is often resource-intensive and time-consuming. While some efforts have been made to explore the synthesis of scene text images, a notable gap remains between syn-thetic and authentic data. In this paper, we introduce a novel method that utilizes Neural Radiance Fields (NeRF) to model real-world scenes and emulate the data collection process by rendering images from diverse camera per-spectives, enriching the variability and realism of the synthesized data. A semi-supervised learning framework is proposed to categorize semantic regions within 3D scenes, ensuring consistent labeling of text regions across various viewpoints. Our method also models the pose, and view-dependent appearance of text regions, thereby offering precise control over camera poses and significantly improving the realism of text insertion and editing within scenes. Employing our technique on real-world scenes has led to the creation of a novel scene text image dataset (https://github.com/cuijl-ai/TextNeRF). Compared to other existing benchmarks, the proposed dataset is distinctive in providing not only standard annotations such as bounding boxes and transcriptions but also the information of 3D pose attributes for text regions, enabling a more detailed evaluation of the robustness of text detection algorithms. Through extensive experiments, we demonstrate the effectiveness of our proposed method in enhancing the performance of scene text detectors.

Although Vision Transformers (ViTs) have recently advanced computer vision tasks significantly, an important real-world problem was overlooked: adapting to variable input resolutions. Typically, images are resized to a fixed resolution, such as 224x224, for efficiency during training and inference. However, uniform input size conflicts with real-world scenarios where images naturally vary in resolution. Modifying the preset resolution of a model may severely degrade the performance. In this work, we propose to enhance the model adaptability to resolution variation by optimizing the patch embedding. The proposed method, called Multi-Scale Patch Embedding (MSPE), substitutes the standard patch embedding with multiple variable-sized patch kernels and selects the best parameters for different resolutions, eliminating the need to resize the original image. Our method does not require high-cost training or modifications to other parts, making it easy to apply to most ViT models. Experiments in image classification, segmentation, and detection tasks demonstrate the effectiveness of MSPE, yielding superior performance on low-resolution inputs and performing comparably on high-resolution inputs with existing methods.

Constantly discovering novel concepts is crucial in evolving environments. This paper explores the underexplored task of Continual Generalized Category Discovery (C-GCD), which aims to incrementally discover new classes from *unlabeled* data while maintaining the ability to recognize previously learned classes. Although several settings are proposed to study the C-GCD task, they have limitations that do not reflect real-world scenarios. We thus study a more practical C-GCD setting, which includes more new classes to be discovered over a longer period, without storing samples of past classes. In C-GCD, the model is initially trained on labeled data of known classes, followed by multiple incremental stages where the model is fed with unlabeled data containing both old and new classes. The core challenge involves two conflicting objectives: discover new classes and prevent forgetting old ones. We delve into the conflicts and identify that models are susceptible to *prediction bias* and *hardness bias*. To address these issues, we introduce a debiased learning framework, namely **Happy**, characterized by **H**ardness-**a**ware **p**rototype sampling and soft entro**py** regularization. For the *prediction bias*, we first introduce clustering-guided initialization to provide robust features. In addition, we propose soft entropy regularization to assign appropriate probabilities to new classes, which can significantly enhance the clustering performance of new classes. For the *harness bias*, we present the hardness-aware prototype sampling, which can effectively reduce the forgetting issue for previously seen classes, especially for difficult classes. Experimental results demonstrate our method proficiently manages the conflicts of C-GCD and achieves remarkable performance across various datasets, e.g., 7.5% overall gains on ImageNet-100. Our code is publicly available at https://github.com/mashijie1028/Happy-CGCD.

Driving behavior classification is an important direction in the field of social transportation systems and advanced driving assistance system (ADAS), which has attracted more and more attention in recent years. An accurate driving behavior classification algorithm plays a great role in traffic safety, energy saving, and other fields. In this article, we propose a novel vehicle speed and image fusion for driving behavior classification network (SIFDriveNet), which classifies driver behaviors into normal driving, aggressive driving, and drowsy driving. Our method has the following key advantages. First, in the research of driving behavior classification, we are the first to introduce a 2-D image with rich roadside information and convert speeds into a 2-D spectrogram expressing the time–frequency characteristics of speeds through short-time Fourier transform (STFT) while unifying the data space of image information and speed information. Second, we propose a tensor fusion method based on weight decomposition to fully fuse the vectors of the two modalities. This method maps the tensor outer product results to the low-dimensional space through weight decomposition and has a low computational cost while maintaining the fusion effect of the tensor outer product. In addition, we evaluated our model on the public UAH-DriveSet and compared it with the most advanced model. Experimental results show that our model has a better performance, and F1-score is 97.9% on all roads. Especially on the secondary road, our F1-score is 99.4%. Also, our model has strong generalization, and we have reached 99.3% F1 in distracted driving multimodal dataset. In addition, the inference speed reaches 411 FPS, enabling real-time needs. The code is available onhttps://github.com/alu222/SIFDriveNet.

Driving behavior classification is a critical component of social transportation systems and advanced driver assistance systems, and it has gained increasing attention in recent years. Accurate classification algorithms for driving behavior play a significant role in enhancing traffic safety, energy conservation, and related fields. In this article, we propose a novel driving behavior classification network named feature-attention-embedding-based multimodal-fusion driving-behavior-classification network (FMDNet). FMDNet incorporates eight types of data, including acceleration along the x-axis, y-axis, z-axis, roll angle, pitch angle, yaw angle, roadside image, and vehicle speed, to classify driving behavior. To effectively fuse features extracted from different modalities, taking into account their varying importance, we introduce the feature attention embedding-based fusion module (FAEF) as our fusion strategy. This fusion strategy enhances the network's capability to capture meaningful features by incorporating two feature attention embedding units that delve deeper into the interplay between different modes. Furthermore, we provide further validation of the effectiveness of our approach through extensive ablation experiments to investigate and analyze the impact of various modal data on the classification of driving behavior. Our proposed FMDNet achieves state-of-the-art performance on the public UAH-DriveSet dataset, demonstrating its effectiveness with an impressive F1-score of 99.0%. Additionally, the robustness of our model is confirmed on distracted dataset, achieving a remarkable F1-score of 99.7%. The model's outstanding performance on both the UAH-DriveSet dataset and the distracted-dataset highlights its capabilities and potential for real-world applications.https://github.com/Wenzhuo-Liu/FMDNet

Although semantic segmentation methods have made remarkable progress so far, their long inference process limits their use in practical applications. Recently, some two-branch and three-branch real-time segmentation networks have been proposed to improve segmentation accuracy by adding branches to extract spatial or border information. For the design of extracting spatial information branches, preserving high-resolution features or adding segmentation loss to guide spatial branches are commonly used methods to extract spatial information. However, these approaches are not the most efficient. To solve the problem, we design the spatial information extraction branch as an AutoEncoder structure, which allows us to extract the spatial structure and features of the image during the encoding and decoding process of the AutoEncoder. Border, semantic and spatial information are all helpful for segmentation tasks, and efficiently fusing these three kinds of information can obtain better feature representation compared to the fusion of two types of information in the dual-branch network. However, existing three-branch networks have yet to explore this aspect deeply. Therefore, this paper designs a new three-branch network based on this starting point. In addition, we also propose a feature fusion module called the Unified Multi-Feature Fusion module (UMF), which can fuse multiple features efficiently. Our method achieves a state-of-the-art trade-off between inference speed and accuracy on the Cityscapes, CamVid, and NightCity datasets. Specifically, BSSNet-T achieves 78.8% mIoU at 115.8 FPS on the Cityscapes dataset, 79.5% mIoU at 170.8 FPS on the CamVid dataset, and 52.6% mIoU at 172.3 FPS on the NightCity dataset. Code is available at https://github.com/SXQ-STUDY/BSSNet.

Since impervious surface is an important indicator of urbanization level and ecological environment, studying the spatio-temporal differentiation of impervious surface has become an urgent issue in urban sustainable development research. Different from previous impervious surface monitoring at intervals of several years, this paper proposed a new spatio-temporal differentiation method of impervious surface on an annual time scale in the Beijing-Tianjin-Hebei urban agglomeration (BTH) from 2000 to 2019. The results show that impervious surface in the BTH has increased steadily, from 16,743.78 km² in 2000 to 28,270.06 km² in 2019. The growth trends of impervious surface are mostly characterized by outward expansion centered on the urban area, and there are obvious differences in the impervious surface expansion of cities within BTH due to different urban patterns and development strategies. These results can provide references for the urban development and planning of national-level urban agglomerations.

This paper introduces Deep Statistical Solvers (DSS), a new class of trainable solvers for optimization problems, arising e.g., from system simulations. The key idea is to learn a solver that generalizes to a given distribution of problem instances. This is achieved by directly using as loss the objective function of the problem, as opposed to most previous Machine Learning based approaches, which mimic the solutions attained by an existing solver. Though both types of approaches outperform classical solvers with respect to speed for a given accuracy, a distinctive advantage of DSS is that they can be trained without a training set of sample solutions. Focusing on use cases of systems of interacting and interchangeable entities (e.g. molecular dynamics, power systems, discretized PDEs), the proposed approach is instantiated within a class of Graph Neural Networks. Under sufficient conditions, we prove that the corresponding set of functions contains approximations to any arbitrary precision of the actual solution of the optimization problem. The proposed approach is experimentally validated on large linear problems, demonstrating super-generalisation properties; And on AC power grid simulations, on which the predictions of the trained model have a correlation higher than 99.99% with the outputs of the classical Newton-Raphson method (known for its accuracy), while being 2 to 3 orders of magnitude faster.

As the large-scale integration of photovoltaic (PV) power station, a higher requirement is put forward by power grid analysis on the accuracy of PV station model. For the model of photovoltaic system is complex, and it requires solving large-scale model equation, which is not conductive for simulation and big data analysis when a large-scale PV power station connected to grid. This paper focuses on the same type of photovoltaic power generation, assuming that the photovoltaic power plant is composed of multiple photovoltaic power generation units connected by the collector line, the photovoltaic power plant can be modeled to consider the operation of similar power generation unit grouped together, so that reduce the simulation scale. This paper points out that the dynamic clustering equivalence strategy can be classified into two cases. During the fault and fault-over, the transient condition is required to be grouped when the active power ramp recovery control module is operating, and the another single-unit equivalent model can be performed when the module is ignoring or not running into a non-fault condition, which is the second condition. Simulation results show that the proposed method has good adaptability for irradiance disturbance, different fault time.

As the development and building of 220kV and 500kV Unified Power Flow Controller (UPFC) in Jiangsu grid of China, a higher requirement is put forward by power grid analysis on the accuracy of modelling and simulation. UPFC has different characteristics under different control strategies, and Modular Multilevel Converters (MMC) simulation is a difficult task. In order to study its dynamic characteristics, an electromagnetic model of UPFC based on MMC was proposed, which could simulate its different control strategies and relative dynamic characteristics. The proposed model was realized in PSD-PSModel program and verified using IEEE39 case. The simulation results show that the proposed UPFC model could provide effective tool for analysing the UPFC's dynamic characteristics.