Published in Association for Computational Linguistics (ACL), 2026 Top 1% Overall Assessment
We propose ActionVLM, a vision-language framework for temporal action localization that uses Language Advantage to adaptively weight language, mitigating language shortcuts and grounding localization in visual evidence.
Under review, 2026
SemVID is a training-free VTG token pruning framework that preserves both boundary-critical evidence and cross-frame reasoning.
Published in Findings of the Association for Computational Linguistics: ACL, 2025
A novel fine-tuning framework to automatically synthesize training data tailored for rejecting the questions exceeds the knowledge without compromising on other tasks.
Under review, 2026
This paper proposes a new input paradigm for mmWave-based human pose estimation, which models human as an Gaussian ensemble enriched with electromagnetic and kinematic parameters.
Published in IEEE International Conference on Multimedia and Expo (ICME), 2026
We propose a lightweight mmWave pose estimation framework that exploits physical priors to reduce parameters by 55.7%–88.9% while maintaining competitive accuracy and enabling real-time Raspberry Pi deployment.
Published in Engineering Applications of Artificial Intelligence, 2025
Decouples redundant and missing features through auxiliary decoders for more effective salient object detection.
Published in IEEE International Conference on Pattern Recognition and Machine Learning (PRML), 2023
A Gaussian mixture model with built-in DTW for clustering variable-length time-series without dimensional explosion.
Under review, 2026
The first automatic faithfulness evaluation metric specifically designed for scientific image tasks.
Published in IEEE Transactions on Circuits and Systems for Video Technology (TCSVT), 2026
Generates multimodal knowledge clues as an image to strengthen image-sequence reasoning with assisted visual input.