@article{videolanguageunderstandingasurveyfrom,
  title = {Video-Language Understanding: A Survey from Model Architecture, Model Training, and Data Perspectives},
  author = {Thong Nguyen and Yi Bin and Junbin Xiao and Leigang Qu and Yicong Li and Jay Zhangjie Wu and Cong-Duy Nguyen and See-Kiong Ng and Luu Anh Tuan},
  year = {2024},
  eprint = {2406.05615},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2406.05615v2},
}