Neural Radiance Fields (NeRF) have gained significant attention due to their precise reconstruction and rapid inference capabilities, making them highly promising for applications in virtual reality and gaming. However, extending NeRF's capabilities to dynamic scenes remains underexplored, particularly in ensuring consistent and coherent reconstructions across space, time, and viewing angles. To address this challenge, we propose Scale-NeRF, a novel approach that organizes the training of dynamic NeRFs as a progressive, scale-based refinement process, grounded in hierarchical Bayesian theory. Scale-NeRF begins by reconstructing the radiance fields using coarse, large-scale frames and iteratively refines them with progressively smaller-scale frames. This hierarchical strategy, combined with a corresponding sampling approach and a newly introduced structural loss, ensures consistency and integrity throughout the reconstruction process. Experiments on public datasets validate the superiority of Scale-NeRF over traditional methods, especially in terms of the proposed metrics evaluating spatial, angular, and temporal consistency. Furthermore, Scale-NeRF demonstrates excellent dynamic reconstruction capabilities with real-time rendering, offering a significant advancement for applications demanding both high fidelity and real-time performance.

In recent years, significant advancements have been made in contrastive self-supervised learning for graphs. However, most of the existing methods start from the feature level and ignore the structural information. In this work, we propose a graph representation learning based on edge structure contrastive self-supervised framework (SCGRL), which leverages a novel edge-structure-based "masked edges vs. complementary edges" instance pairs to fully utilize the topological information of the graph, and attempts to reconstruct the original graph using the visible graph structure. In the feature processing, normal coded features are constrained with the coded features without gradient updating to enhance the encoder’s prediction ability for the masked representation. In addition, boundary losses are designed to ensure that the model can accurately distinguish between different instance pairs. We conduct extensive experiments on various benchmark datasets to demonstrate that SCGRL outperforms the state-of-the-art in different downstream tasks, especially link prediction.

Time series analysis plays a fundamental role in revealing data evolution, trends, and cyclical patterns. However, existing studies often fail to effectively address the dynamic dependencies between variables in multidimensional time series and the temporal evolution patterns within variables, thereby limiting the effectiveness of complex time series feature analysis. In this paper, we propose a dual-branch frequency-time interactive representation learning model (FreqTime) that captures the correlations between variables and the temporal dependencies within variables through a collaborative architecture in the time domain and frequency domain. The time domain branch uses an inverse Transformer architecture to model cross-variable interactions, while the frequency domain branch utilizes multi-scale gated convolutions to capture features and map them back to the time domain. Finally, global representations are obtained by interactively fusing the representations learned from the two branches in the time domain. Experiments demonstrate that FreqTime achieves state-of-the-art performance on long sequence prediction, classification, and anomaly detection tasks, and exhibits strong robustness in noisy environments.

Queue length is one of the most commonly used indicators to evaluate traffic operation at signalized intersections. Many studies aim to estimate queue length using trajectory data, but this remains a challenge with spatially sparse trajectories. This paper introduces a probabilistic-based method for cycle-by-cycle estimation of queue length distribution and the point estimate in closed form using connected vehicle (CV) trajectories. The method is applicable to isolated signalized intersections with under-saturated traffic. It works well in a low CV penetration rate environment by exploiting the trajectories of both queued and non-queued CVs. Vehicle arrival rates and CV penetration rates are first estimated to capture the vehicle arrival pattern within a time of day (TOD) using the maximum likelihood estimation approach. The closed forms of the probability distribution of queue length are derived for each cycle, which are attractive for applications such as adaptive signal timing considering traffic uncertainty. The queue length with the highest probability is taken as the point estimate. Numerical and empirical studies demonstrate that the proposed method outperforms the benchmark method in estimating CV penetration rates, particularly at low penetration rates. In term of queue length estimation, the proposed method is also superior to existing methods in most scenarios and is especially effective for cycles with observed trajectories. Sensitivity analysis reveals that the method is robust to different demand levels, signal timings, and variability in arrival patterns with under-saturated traffic. Additionally, the suggested requirements for the number of collected trajectories are investigated to provide practical guidance.

Visual bird’s eye view (BEV) semantic segmentation helps autonomous vehicles understand the surrounding environment only from front-view (FV) images, including static elements (e.g., roads) and dynamic elements (e.g., vehicles, pedestrians). However, the high cost of annotation procedures of full-supervised methods limits the capability of the visual BEV semantic segmentation, which usually needs HD maps, 3D object bounding boxes, and camera extrinsic matrixes. In this paper, we present a novel semi-supervised framework for visual BEV semantic segmentation to boost performance by exploiting unlabeled images during the training. A consistency loss that makes full use of unlabeled data is then proposed to constrain the model on not only semantic prediction but also the BEV feature. Furthermore, we propose a novel and effective data augmentation method named conjoint rotation which reasonably augments the dataset while maintaining the geometric relationship between the FV images and the BEV semantic segmentation. Extensive experiments on the nuScenes dataset show that our semi-supervised framework can effectively improve prediction accuracy. To the best of our knowledge, this is the first work that explores improving visual BEV semantic segmentation performance using unlabeled data. The code is available at https://github.com/Junyu-Z/Semi-BEVseg.

Semantic scene completion (SSC) aims to predict the semantic occupancy of each voxel in the entire 3D scene from limited observations, which is an emerging and critical task for autonomous driving. Recently, many studies have turned to camera-based SSC solutions due to the richer visual cues and cost-effectiveness of cameras. However, existing methods usually rely on sophisticated and heavy 3D models to process the lifted 3D features directly, which are not discriminative enough for clear segmentation boundaries. In this paper, we adopt the dense-sparse-dense design and propose a one-stage camera-based SSC framework, termed SGN, to propagate semantics from the semantic-aware seed voxels to the whole scene based on spatial geometry cues. Firstly, to exploit depth-aware context and dynamically select sparse seed voxels, we redesign the sparse voxel proposal network to process points generated by depth prediction directly with the coarse-to-fine paradigm. Furthermore, by designing hybrid guidance (sparse semantic and geometry guidance) and effective voxel aggregation for spatial geometry cues, we enhance the feature separation between different categories and expedite the convergence of semantic propagation. Finally, we devise the multi-scale semantic propagation module for flexible receptive fields while reducing the computation resources. Extensive experimental results on the SemanticKITTI and SSCBench-KITTI-360 datasets demonstrate the superiority of our SGN over existing state-of-the-art methods. And even our lightweight version SGN-L achieves notable scores of 14.80% mIoU and 45.45% IoU on SeamnticKITTI validation with only 12.5 M parameters and 7.16 G training memory. Code is available at https://github.com/Jieqianyu/SGN.

The great potential of unsupervised monocular depth estimation has been demonstrated by many works due to low annotation cost and impressive accuracy comparable to supervised methods. To further improve the performance, recent works mainly focus on designing more complex network structures and exploiting extra supervised information, e.g., semantic segmentation. These methods optimize the models by exploiting the reconstructed relationship between the target and reference images in varying degrees. However, previous methods prove that this image reconstruction optimization is prone to get trapped in local minima. In this paper, our core idea is to guide the optimization with prior knowledge from pretrained Flow-Net. And we show that the bottleneck of unsupervised monocular depth estimation can be broken with our simple but effective framework named FG-Depth. In particular, we propose (i) a flow distillation loss to replace the typical photometric loss that limits the capacity of the model and (ii) a prior flow based mask to remove invalid pixels that bring the noise in training loss. Extensive experiments demonstrate the effectiveness of each component, and our approach achieves state-of-the-art results on both KITTI and NYU-Depth-v2 datasets.

Improving the generalization of neural networks under domain shift is an important and challenging task in computer vision. Obtaining an invariant representation across domains is a benchmark method in the literature. In this paper, we propose an invariant and sufficient supervised representation learning (ISSRL) approach to learn a domain invariant representation which is also preserving information used for downstream tasks. To this end, we formulate ISSRL by finding a nonlinear map$\boldsymbol{g}$such that$Y\perp X\vert \boldsymbol{g}(X)$and$(Y,\boldsymbol{g}(X))\perp D$at the population level, where D is the label of the domains and$(X, Y)$is the paired data sampled from domains with label. We use distance correlation to characterize the (conditional) independence. At the sample level, we construct a novel loss function through an unbiased empirical version of distance correlation. We train the representation map by parameterizing it with deep neural networks. Both simulation study and real data evaluation show that ISSRL outperforms the state-of-the-art on out-of-distribution generalization. The PyTorch code for ISSRL is available at https://github.com/CaC033/ISSRL.

An event camera is a novel vision sensor that can capture per-pixel brightness changes and output a stream of asynchronous “events”. It has advantages over conventional cameras in those scenes with high-speed motions and challenging lighting conditions because of the high temporal resolution, high dynamic range, low bandwidth, low power consumption, and no motion blur. Therefore, several supervised monocular depth estimation from events is proposed to address scenes difficult for conventional cameras. However, depth annotation is costly and time-consuming. In this paper, to lower the annotation cost, we propose a self-supervised event-based monocular depth estimation framework named EMoDepth. EMoDepth constrains the training process using the cross-modal consistency from intensity frames that are aligned with events in the pixel coordinate. Moreover, in inference, only events are used for monocular depth prediction. Additionally, we design a multi-scale skip-connection architecture to effectively fuse features for depth estimation while maintaining high inference speed. Experiments on MVSEC and DSEC datasets demonstrate that our contributions are effective and that the accuracy can outperform existing supervised event-based and unsupervised frame-based methods.

Real aperture super-resolution (RAS) technology is widely used in the field of radar forward-looking imaging. However, traditional RAS technology is based on the space-to-ground scanning mode. The essence of this technology is azimuth (angle) super-resolution, which is a one-dimensional super-resolution technology. In our work, we consider applying RAS technology to the space-to-space scanning. In this mode, we regard the echo of each range slice as the convolution of the target scattering coefficient distribution and the antenna pattern function. Its essence is azimuth and pitch super-resolution, which is a two-dimensional super-resolution technology. Finally, a reasonable objective function is constructed under the framework of regularization, and the ADMM solver is used to achieve two-dimensional super-resolution imaging. Simulations will prove the effectiveness of the proposed two-dimensional super-resolution algorithm.

Multistatic synthetic aperture radar (SAR) can adopt flexible topology structures to accomplish different missions. When we aim to coherently fuse multiple measurements of receivers, the topology structure of multi static SAR is the key to affect the imaging quality. In this paper, a topology design method based on wavenumber spectrum generation is proposed. The wavenumber spectrum distribution forms the dependency relationship between the imaging quality and topology structures. Based on the analysis of the kernel wavenumber spectrum distribution, the wavenumber spectrum generation is proposed to improve the spatial resolution. Using the generated wavenumber spectrum, the topology structure can be designed accurately. The proposed method effectively enhances the imaging resolution of multi static SAR at a low time cost. Simulation results verify the validity of the proposed method.

Due to the dynamically changing topology of Internet of Vehicles (IoV), it is a challenging issue to achieve efficient data dissemination in IoV. This paper considers strongly connected IoV with a number of heterogenous vehicular nodes to disseminate information and studies distributed replication-based data dissemination algorithms to improve the performance of data dissemination. Accordingly, two data replication algorithms, a deterministic algorithm and a distributed randomised algorithm, are proposed. In the proposed algorithms, the number of message copies spread in the network is limited and the network will be balanced after a series of average operations among the nodes. The number of communication stages needed for network balance shows the complexity of network convergence as well as network convergence speed. It is proved that the network can achieve a balanced status after a finite number of communication stages. Meanwhile, the upper and lower bounds of the time complexity are derived when the distributed randomised algorithm is applied. Detailed mathematical results show that the network can be balanced quickly in complete graph; thus highly efficient data dissemination can be guaranteed in dense IoV. Simulation results present that the proposed randomised algorithm outperforms the present schemes in terms of transmissions and dissemination delay.

In this paper, we investigate the problem of optimal road side units (RSUs) placement in Vehicular Ad Hoc Network (VANET) on a highway, which enables the VANET maintain a good connectivity. Our goal is to find out minimal number of road side units, such that the vehicles could communicate with RSUs. These road side units are connected by wire. We develop a randomized algorithm to deploy road side units in the VANET. It gives an approximation to the optimal distance to guarantee the information can be passed to RSUs from the accident site via the VANET. Simulations are conducted to show the performance of our proposed method.