#cross-attention
17 papers · 1 filter
BlindPSNR: A No-Reference Fidelity Predictor for Low-Light Image Enhancement
Mingzhe Lyu, Jinqiang Cui, Hong Zhang
The paper proposes BlindPSNR, a lightweight no‑reference network that predicts PSNR for low‑light image enhancement by fusing the enhanced image with the original low‑light input v…
Drawing-Recode: Annotation Grounding for Parametric CAD Code Generation from Raster 2D CAD Drawings
Mingi Kim, Yongjun Kim, Hyungki Kim
Drawing-Recode is a system that converts raster 2D CAD drawings into parametric CAD code by extracting geometric features, recognizing textual annotations, grounding the annotation…
Vision-TL-Action: Neuro-Symbolic Trajectory Generation from Visual Observations and Temporal Logic
Zezhi Liu, Zhiwei Zheng, Hanqian Luo +3
The paper introduces Vision-TL-Action, a neuro‑symbolic system that generates robot action trajectories directly from multi‑view images and a coordinate‑free temporal‑logic goal re…
FoMoVLA: Bridging Visual Foresight and Motion Guidance for Vision-Language-Action Models
Wei Li, Peijin Jia, Yuan Ma +9
FoMoVLA enhances vision-language-action models by jointly predicting future visual features and tracking sparse 2D points, providing both goal states and motion paths to improve co…
LPCAN: Lightweight Pyramid Cross-Attention Network for Rail Surface Defect Detection Using RGB-D Data
Jackie Alex, Guoqiang Huan
The paper introduces a lightweight pyramid cross‑attention network (LPCANet) that uses RGB‑D data to detect rail surface defects efficiently, achieving high accuracy with low compu…
Thresholded Cross-Attention for Reliable Intensity-Chromaticity Fusion in Low-Light Image Enhancement
Yanyi Wu, Xu Zhang, Junkai Chen +6
The paper introduces TCA-Net, which uses a thresholded cross-attention mechanism to more reliably fuse intensity and chromaticity information for low-light image enhancement, impro…