Previous camera self-calibration methods have exhibited certain notable shortcomings. On the one hand, they either exclusively emphasized scene cues or solely focused on vehicle-related cues, resulting in a lack of adaptability to diverse scenarios and a limited number of effective features. Furthermore, these methods either solely utilized geometric features within traffic scenes or exclusively extracted semantic information, failing to comprehensively consider both aspects. This limited the comprehensive feature extraction from scenes, ultimately leading to a decrease in calibration accuracy. Additionally, conventional vanishing point-based self-calibration methods often required the design of additional edge-background models and manual parameter tuning, thereby increasing operational complexity and the potential for errors. Given these observed limitations, and in order to address these challenges, we propose an innovative roadside camera self-calibration model based on the Transformer architecture. This model possesses a unique capability to simultaneously learn scene features and vehicle features within traffic scenarios while considering both geometric and semantic information. Through this approach, our model can overcome the constraints of prior methods, enhancing calibration accuracy and robustness while reducing operational complexity and the potential for errors. Our method outperforms existing approaches on both real-world dataset scenarios and publicly available datasets, demonstrating the effectiveness of our approach.
Previous camera self-calibration methods have exhibited certain notable shortcomings. On one hand, they either exclusively emphasized scene cues or solely focused on vehicle-related cues, resulting in a lack of adaptability to diverse scenarios and a limited number of effective features. Furthermore, these methods either solely utilized geometric features within traffic scenes or exclusively extracted semantic information, failing to comprehensively consider both aspects. This limited the comprehensive feature extraction from scenes, ultimately leading to a decrease in calibration accuracy. Additionally, conventional vanishing point-based self-calibration methods often required the design of additional edge-background models and manual parameter tuning, thereby increasing operational complexity and the potential for errors. Given these observed limitations, and in order to address these challenges, we propose an innovative roadside camera self-calibration model based on the Transformer architecture. This model possesses a unique capability to simultaneously learn scene features and vehicle features within traffic scenarios while considering both geometric and semantic information. Through this approach, our model can overcome the constraints of prior methods, enhancing calibration accuracy and robustness while reducing operational complexity and the potential for errors. Our method outperforms existing approaches on both real-world dataset scenarios and publicly available datasets, demonstrating the effectiveness of our approach.
14 November 2025 in IEEE Transactions on Intelligent Transportation Systems
Vehicle-to-Everything (V2X) collaborative perception bolsters the performance of autonomous vehicles by overcoming occlusion challenges and expanding their sensing range. However, current methodologies frequently overlook the intrinsic uncertainties associated with localization inaccuracies, asynchronous measurements, and diverse agent models. Such uncertainties may weaken fusion reliability, leading to performance worse than single-vehicle perception. To address this pivotal challenge, we introduce BELT-Fusion, a cohesive probabilistic framework tailored for reliable V2X late fusion. Our framework offers two notable advantages. The first is explicit agent-level uncertainty modeling, where classification uncertainty is captured via evidential deep learning and regression uncertainty via Bayesian neural networks. This capability allows task-specific reliability assessments to be effortlessly incorporated into existing object detectors. Second, our framework introduces an uncertainty-aware adaptive fusion representation. This representation dynamically guides object selection and weight allocation based on measurable fusion-level uncertainty, ensuring reliable fusion results without retraining and enabling plug-and-play functionality. To validate BELT-Fusion’s efficacy, we conducted evaluations focusing on 3D object detection in both real-world and simulated scenarios using the OPV2V and DAIR-V2X datasets. BELT-Fusion improved AP@0.7 by 7.16% in noisy settings and 3.84% under ideal conditions over uncertainty-agnostic baselines, demonstrating its robustness under challenging and noisy conditions. Our code will be available at https://github.com/ZhiguoZhao/BELT-Fusion
Spatial synchronization in roadside scenarios is essential for integrating data from multiple sensors at different locations. Current methods using cascading spatial transformation (CST) often lead to cumulative errors in large-scale deployments. Manual camera calibration is insufficient and requires extensive manual work, and existing methods are limited to controlled or single-view scenarios. To address these challenges, our research introduces a parallel spatial transformation (PST)-based framework for large-scale, multi-view, multi-sensor scenarios. PST parallelizes sensor coordinate system transformation, reducing cumulative errors. We incorporate deep learning for precise roadside monocular global localization, reducing manual work. Additionally, we use geolocation cues and an optimization algorithm for improved synchronization accuracy. Our framework has been tested in real-world scenarios, outperforming CST-based methods. It significantly enhances large-scale roadside multi-perspective, multi-sensor spatial synchronization, reducing deployment costs.