arXiv subjects

Hui Sheng

Publications and source records attributed to Hui Sheng.

3 recordsLinked to original sources

arXiv2026

PolyBuild: An End-to-End Method for Polygonal Building Contour Extraction from High-Resolution Remote Sensing Images

Extracting building polygon contours from high-resolution remote sensing images is a fundamental task for various mapping applications. However, the presence of varying imaging conditions and complex building structures, makes automatic contour extraction extremely challenging. Mainstream approaches for building extraction often rely on pixel-level segmentation followed by multiple post-processing steps to produce building contour, which can be computationally intensive and prone to errors. In this paper, we propose an end-to-end method named PolyBuild, which can directly extract building vector polygons from high-resolution remote sensing images without the need for any post-processing operations. The proposed method leverages two primary modules: an Initial Contour Generation Module (ICGM) and a Contour Optimization Module (COM). The ICGM is designed to generate an initial building contour by utilizing concatenated sub-region center features for each building instance. It performs simultaneous object detection and initial contour extraction by generating bounding boxes and using the center features of four sub-regions to represent each building. The Contour Optimization Module (COM) further refines the generated building contours by iteratively integrating Convolutional Neural Network (CNN) features and contour positional information in a Transformer-based decoder. The hybrid CNN-Transformer architecture effectively captures both local and global spatial relationships within the building contour, ensuring high-quality boundary delineation. Extensive experiments are conducted on three building datasets to evaluate the performance of PolyBuild. The results demonstrate that PolyBuild significantly outperforms state-of-the-art methods, including mask-based and contour-based approaches.

cs.CV↗

arXiv2024

Know in AdVance: Linear-Complexity Forecasting of Ad Campaign Performance with Evolving User Interest

Real-time Bidding (RTB) advertisers wish to \textit{know in advance} the expected cost and yield of ad campaigns to avoid trial-and-error expenses. However, Campaign Performance Forecasting (CPF), a sequence modeling task involving tens of thousands of ad auctions, poses challenges of evolving user interest, auction representation, and long context, making coarse-grained and static-modeling methods sub-optimal. We propose \textit{AdVance}, a time-aware framework that integrates local auction-level and global campaign-level modeling. User preference and fatigue are disentangled using a time-positioned sequence of clicked items and a concise vector of all displayed items. Cross-attention, conditioned on the fatigue vector, captures the dynamics of user interest toward each candidate ad. Bidders compete with each other, presenting a complete graph similar to the self-attention mechanism. Hence, we employ a Transformer Encoder to compress each auction into embedding by solving auxiliary tasks. These sequential embeddings are then summarized by a conditional state space model (SSM) to comprehend long-range dependencies while maintaining global linear complexity. Considering the irregular time intervals between auctions, we make SSM's parameters dependent on the current auction embedding and the time interval. We further condition SSM's global predictions on the accumulation of local results. Extensive evaluations and ablation studies demonstrate its superiority over state-of-the-art methods. AdVance has been deployed on the Tencent Advertising platform, and A/B tests show a remarkable 4.5\% uplift in Average Revenue per User (ARPU).

cs.IR↗

arXiv2021

Plasma response to resonant magnetic perturbations near rotation zero-crossing in low torque plasmas

Plasma response to resonant magnetic perturbations (RMPs) near the pedestal top is crucial for accessing edge localized modes (ELMs) suppression in tokamaks. Since radial location of rotation zero-crossing plays a key role in determining the threshold for field penetration of RMP, plasma response may be different in low input torque plasmas. In this work, the linear MHD code MARS-F is applied to reveal the dependence of plasma response to RMP on rotation zero-crossing by a scan of rotation profiles based on an EAST equilibrium. It is shown that the plasma response is enhanced when zero-crossing occurs near rational surfaces. The dependence of plasma response on the location of rotation zero-crossing is well fitted by a double Gaussian, indicating two effects in this enhancement. One is induced by rotation screening effect shown as a wide base (with a width around 10-20 krad/s), and the other is related to resistive singular layer effect characterized by a localized peak (with a width around 3-4 krad/s). The width of the peak scales with the resistive singular layer width. The plasma displacement suggests the response is tearing like when zero-crossing is within the singular layer, while it is kink like when zero-crossing is far from the layer. The enhancement of magnetic islands width at the peak is only around a factor of two, when the absolute value of local rotation is not larger than 10-20 krad/s. It is further confirmed in a modeling of plasma response in an EAST ELM suppression discharge. Though there is a zero-crossing in $E\times B$ rotation but not in electron perpendicular rotation, no significant difference in plasma response is obtained using these two rotation profiles. This suggests that the rotation near pedestal top should not be far away from zero but may not be necessary to have zero-crossing for accessing ELM suppression.

physics.plasm-ph↗