JYJun Yan

Jun Yan

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Jun Yan has a total of 33 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Sparse Query Dense: Enhancing 3D Object Detection with Pseudo Points

    by Yujian Mo, Yan Wu, Junqiao Zhao, Zhenjie Hou, Weiquan Huang, Yinghao Hu, Jijun Wang, Jun Yan

    Current LiDAR-only 3D detection methods are limited by the sparsity of point clouds. The previous method used pseudo points generated by depth completion to supplement the LiDAR point cloud, but the pseudo points sampling process was complex, and the distribution of pseudo points was uneven. Meanwhile, due to the imprecision of depth completion, the pseudo points suffer from noise and local structural ambiguity, which limit the further improvement of detection accuracy. This paper presents SQDNet, a novel framework designed to address these challenges. SQDNet incorporates two key components: the SQD, which achieves sparse-to-dense matching via grid position indices, allowing for rapid sampling of large-scale pseudo points on the dense depth map directly, thus streamlining the data preprocessing pipeline. And use the density of LiDAR points within these grids to alleviate the uneven distribution and noise problems of pseudo points. Meanwhile, the sparse 3D Backbone is designed to capture long-distance dependencies, thereby improving voxel feature extraction and mitigating local structural blur in pseudo points. The experimental results validate the effectiveness of SQD and achieve considerable detection performance for difficult-to-detect instances on the KITTI test.


  • Open Access

    An uncertainty-aware domain adaptive semantic segmentation framework

    by Huilin Yin, Pengyu Wang, Boyu Liu, Jun Yan

    Semantic segmentation is significant to realize the scene understanding of autonomous driving. Due to the lack of annotated real-world data, the technology of domain adaptation is applied so that the model is trained on the synthetic data and inferred on the real data. However, this domain gap leads to aleatoric and epistemic uncertainty. These uncertainties link to the potential safety issue of autonomous driving in normal weather and adverse weather. In this study, we explore the scientific problem that has received sparse attention previously. We postulate that the Dual Attention module can mitigate the uncertainty in the task of semantic segmentation and provide some empirical study to validate it. Furthermore, the utilization of Kullback-Leibler divergence (KL divergence) helps the estimation of aleatoric uncertainty and boosts the robustness of the segmentation model. Our empirical study on the diverse datasets of semantic segmentation demonstrates the effectiveness of our method in normal and adverse weather. Our code is available at: https://github.com/liubo629/Seg-Uncertainty-dual-attention.


  • Open Access

    PCQA: A Strong Baseline for AIGC Quality Assessment Based on Prompt Condition

    by Xi Fang, Weigang Wang, Xiaoxin Lv, Jun Yan

    The development of Large Language Models (LLM) and Diffusion Models brings the boom of Artificial Intelligence Generated Content (AIGC). It is essential to build an effective quality assessment framework to provide a quantifiable evaluation of different images or videos based on the AIGC technologies. The content generated by AIGC methods is driven by the crafted prompts. Therefore, it is intuitive that the prompts can also serve as the foundation of the AIGC quality assessment. This study proposes an effective AIGC quality assessment (QA) framework. First, we propose a hybrid prompt encoding method based on a dual-source CLIP (Contrastive Language-Image Pre-Training) text encoder to understand and respond to the prompt conditions. Second, we propose an ensemble-based feature mixer module to effectively blend the adapted prompt and vision features. The empirical study practices in two datasets: AIGIQA-20K (AI-Generated Image Quality Assessment database) and T2VQA-DB (Text-to-Video Quality Assessment DataBase), which validates the effectiveness of our proposed method: Prompt Condition Quality Assessment (PCQA). Our proposed simple and feasible framework may promote research development in the multimodal generation field.

  • Open Access

    SA-Attack: Speed-adaptive stealthy adversarial attack on trajectory prediction

    by Huilin Yin, Jiaxiang Li, Pengju Zhen, Jun Yan

    Trajectory prediction is critical for the safe planning and navigation of automated vehicles. The trajectory prediction models based on the neural networks are vulnerable to adversarial attacks. Previous attack methods have achieved high attack success rates but overlook the adaptability to realistic scenarios and the concealment of the deceits. To address this problem, we propose a speed-adaptive stealthy adversarial attack method named SA-Attack. This method searches the sensitive region of trajectory prediction models and generates the adversarial trajectories by using the vehicle-following method and incorporating information about forthcoming trajectories. Our method has the ability to adapt to different speed scenarios by reconstructing the trajectory from scratch. Fusing future trajectory trends and curvature constraints can guarantee the smoothness of adversarial trajectories, further ensuring the stealthiness of attacks. The empirical study on the datasets of nuScenes and Apolloscape demonstrates the attack performance of our proposed method. Finally, we also demonstrate the adaptability and stealthiness of SA-Attack for different speed scenarios. Our code is available at the repository: https://github.com/eclipse-bot/SA-Attack.

  • BEVSOC: Self-Supervised Contrastive Learning for Calibration-Free BEV 3-D Object Detection

    by Yongqing Chen, Nanyu Li, Dandan Zhu, Charles C. Zhou, Zhuhua Hu, Yong Bai, Jun Yan

    3D object detection based on multi-view cameras and bird’s-eye view (BEV) representation is a key task for autonomous driving, as it enables the perception systems to understand the surrounding scenes. However, most existing BEV representation methods rely on the projection matrix of camera intrinsic and extrinsic parameters, which requires a complex and time-consuming calibration process that may introduce errors and degrade the detection performance. Moreover, the calibration results may vary due to environmental changes and affect the stability of the detection system. To address this problem, we propose a calibration-free 3D object detection method that leverages a group-equivariant convolutional network to extract features from multi-view images and a projection network module to learn the implicit 3D-to-2D projection relationship for obtaining BEV representation. Furthermore, we employ contrastive learning to pre-train the projection network module without using manually annotated data. By exploiting the multi-view camera data through contrastive learning, our proposed method eliminates the need for tedious calibration, avoids calibration errors, and reduces the dependence on a large amount of annotated data for calibration-free 3D object detection. We evaluate our method on the nuScenes dataset and demonstrate its competitive performance. Our method improves the stability and reliability of 3D object detection in long-term autonomous driving.


  • Loosely Coupled Stereo VINS Based on Point-Line Features Tracking With Feedback Loops

    by Linchuan Zhang, Wei Ye, Jun Yan, Hao Zhang, Johannes Betz, Huilin Yin

    Visual-inertial navigation system (VINS) is a state-of-the-art technology for estimating the motion and position of moving objects, such as drones and robots. Tightly coupled methods have high accuracy but low real-time performance, while loosely coupled systems are more efficient, especially for resource-constrained systems. This study proposes a real-time loosely coupled stereo VINS based on point-line features. Unlike filter-based or optimization-based approaches, the pose estimation process is modeled as a control system. Firstly, as a loosely coupled scheme, the inertial measurement unit (IMU) data is used as the core of the proposed system. Feedback loops are introduced to correct the IMU bias, including a visual-inertial inter-frame bias estimation feedback loop and a versatile quaternion-based filter (VQF) gyroscope bias estimation feedback loop. Secondly, the accelerometer inclination correction based on VQF is introduced in the IMU propagation stage to account for slow inclination drift. To achieve better visual-inertial calibration bias results, line features are introduced in the visual part to improve accuracy and robustness. The scale of the line features is controlled by judging distance and gradient thresholds to improve the real-time performance of the edge drawing lines (EDLines) algorithm. The proposed method is evaluated on the EuRoC MAV dataset and tested on the KITTI dataset. The comparisons with previous methods on benchmark datasets demonstrate the effectiveness of the proposed method in terms of localization accuracy and real-time performance.


  • Enhance Adversarial Robustness via Geodesic Distance

    by Jun Yan, Huilin Yin, Ziming Zhao, Wancheng Ge, Jingfeng Zhang

    Adversarial training is an effective method to improve the model’s adversarial robustness. To realize a considerable tradeoff between clean accuracy and adversarial robustness, surrogate loss minimization can be utilized to regularize for robustness. This study advances beyond previous research efforts by taking a further step. It is informed by two theoretical perspectives. First, adversarial examples are inevitable within a unit sphere surrounding clean data. Second, geodesics can characterize the shortest distance between two points in Riemannian geometry. Accordingly, this study employs geodesic distance as a regularized surrogate loss item to capture the minimal divergence between the distribution of natural examples and the distribution of adversarial examples. This approach yields a more compact upper bound of risk errors than previous studies, which is beneficial for improving adversarial robustness. Based on the theoretical insight, this study proposes a metric of Geodesic Loss and a framework of Geodesic Adversarial Training to boost the adversarial robustness of neural networks. The empirical study on the diverse datasets demonstrates the considerable performance of the proposed method against diverse attacks, including white-box attacks, black-box corruptions, adaptive attacks, and auto attacks. Our code is available at: https://github.com/momo1986/GeodesicAdversarialTraining .


  • Open Access

    Multi-agent Reinforcement Learning for Cooperative Lane Changing of Connected and Autonomous Vehicles in Mixed Traffic

    by Wei Zhou, Dong Chen, Jun Yan, Zhaojian Li, Huilin Yin, Wanchen Ge

    Autonomous driving has attracted significant research interests in the past two decades as it offers many potential benefits, including releasing drivers from exhausting driving and mitigating traffic congestion, among others. Despite promising progress, lane-changing remains a great challenge for autonomous vehicles (AV), especially in mixed and dynamic traffic scenarios. Recently, reinforcement learning (RL), a powerful data-driven control method, has been widely explored for lane-changing decision makings in AVs with encouraging results demonstrated. However, the majority of those studies are focused on a single-vehicle setting, and lane-changing in the context of multiple AVs coexisting with human-driven vehicles (HDVs) have received scarce attention. In this paper, we formulate the lane-changing decision making of multiple AVs in a mixed-traffic highway environment as a multi-agent reinforcement learning (MARL) problem, where each AV makes lane-changing decisions based on the motions of both neighboring AVs and HDVs. Specifically, a multi-agent advantage actor-critic network (MA2C) is developed with a novel local reward design and a parameter sharing scheme. In particular, a multi-objective reward function is proposed to incorporate fuel efficiency, driving comfort, and safety of autonomous driving. Comprehensive experimental results, conducted under three different traffic densities and various levels of human driver aggressiveness, show that our proposed MARL framework consistently outperforms several state-of-the-art benchmarks in terms of efficiency, safety and driver comfort.