works on

From the 1 of 10 linked papers with an AI index.

activity
20242026
collaborators
Showing cs.CVShow all

9 papers · 1 filter

cs.CV2026

Training-Free Decoding of SAM3 Semantic Responses for Cross-Domain Infrastructure Crack Segmentation

Shipeng Liu, Zhanping Song, Liang Zhao +1

The paper introduces Semantic-Edge Response Decoding (SERD), a method that extracts internal semantic responses from the SAM3 vision model and refines them with an edge prior to ac…

cs.CV2026

Rethinking Efficient Crack Segmentation with Task-Aligned Structural-Directional Modeling

Shipeng Liu, Liang Zhao, Dengfeng Chen +1

Recent crack segmentation methods often follow generic semantic segmentation designs, using stronger backbones, hybrid CNN-Transformer-Mamba encoders, and auxiliary enhancement bra…

cs.CV2026

Training-Free Tunnel Defect Inspection and Engineering Interpretation via Visual Recalibration and Entity Reconstruction

Shipeng Liu, Liang Zhao, Dengfeng Chen +1

Tunnel inspection requires outputs that can support defect localization, measurement, severity grading, and engineering documentation. Existing training-free foundation-model pipel…

cs.CV2026

MATTER: Multiscale Attention for Registration Error Regression

Shipeng Liu, Ziliang Xiong, Khac-Hoang Ngo +1

Point cloud registration (PCR) is crucial for many downstream tasks, such as simultaneous localization and mapping (SLAM) and object tracking. This makes detecting and quantifying…

cs.CV2025

Describe-to-Score: A text-guided framework for image complexity assessment

Shipeng Liu, Zhonglin Zhang, Dengfeng Chen +1

Accurately assessing image complexity (IC) is essential for many vision tasks, yet existing approaches rely almost exclusively on visual features and therefore fail to capture the…

cs.CV2025

Continuous Normalizing Flows for Uncertainty-Aware Human Pose Estimation

Shipeng Liu, Ziliang Xiong, Bastian Wandt +1

Human Pose Estimation (HPE) is increasingly important for applications like virtual reality and motion analysis, yet current methods struggle with balancing accuracy, computational…