Skip to content

Latest commit

 

History

History
32 lines (20 loc) · 1.15 KB

README.md

File metadata and controls

32 lines (20 loc) · 1.15 KB
Raw

Layered Memory Network (LMN)

The LMN model ranked 1st place on MovieQA Video+Subtt-based Answering Challenge 2017 (The Joint Video and Language Understanding Workshop, ICCV 2017).

  • The flowchart of Layered Memory Network (LMN).

LMN

  • The framework of Dynamic Subtitle Memory module with update mechanism.

DSM

Train

python mqa_video+subtitle+update+question.py

Paper

Bo Wang, Youjiang Xu, Yahong Han, Richang Hong. "Movie Question Answering: Remembering the Textual Cues for Layered Visual Contents." AAAI, 2018. [Paper]

@inproceedings{Wang2018,
  author    = {Bo Wang and Youjiang Xu and Yahong Han and Richang Hong},
  title     = {Movie Question Answering: Remembering the Textual Cues for Layered Visual Contents},
  booktitle = {AAAI},
  year      = {2018},
}