Published in EMNLP 2025 Oral, 2025
LangDC is a language-aware dynamic token compressor for large video-language models, accepted as an EMNLP 2025 Oral paper.
Recommended citation: Xiangchen Wang, Jinrui Zhang, Teng Wang, Haigang Zhang, and Feng Zheng. (2025). "Seeing More, Saying More: Lightweight Language Experts are Dynamic Video Token Compressors." EMNLP 2025 Oral.
Download Paper
Published in arXiv, 2026
NaVIDA is a lightweight vision-language navigation framework that learns action-grounded visual dynamics with inverse dynamics supervision.
Recommended citation: Weiye Zhu, Zekai Zhang, Xiangchen Wang, Hewei Pan, Teng Wang, Tiantian Geng, Rongtao Xu, and Feng Zheng. (2026). "NaVIDA: Vision-Language Navigation with Inverse Dynamics Augmentation." arXiv:2601.18188.
Download Paper
Published in ACL 2026, 2026
CAPruner prunes 3D scene graphs by combining semantic relevance with spatial proximity to preserve task-critical relations.
Recommended citation: Shengli Zhou, Xiangchen Wang, Guanhua Chen, and Feng Zheng. (2026). "CAPruner: Conceptual-Adjacent Scene Graph Pruner for Enhancing 3D Spatial Reasoning of Large Language Models." ACL 2026.
Download Paper
Published in arXiv, 2026
LiveVLN is a training-free runtime framework that keeps actions continuously available during embodied vision-language navigation.
Recommended citation: Xiangchen Wang, Weiye Zhu, Teng Wang, TianTian Geng, Zekai Zhang, Zhiyuan Qi, Jinyu Yang, and Feng Zheng. (2026). "LiveVLN: Breaking the Stop-and-Go Loop in Vision-Language Navigation." arXiv:2604.19536.
Download Paper