PyTorch 1.6.0 implementation of the following paper.
Demo.py for model test.
opt.py for the setting of hyper-parameters.
Pretrained model weight available at: https://pan.quark.cn/s/bc3ee29dd408
The precomputed region features from the raw images using bottom-up-attention model.
If you find our work is useful for your project or research, please consider citing our paper:
@ARTICLE{ITIN,
author={Zhu, Tong and Li, Leida and Yang, Jufeng and Zhao, Sicheng and Liu, Hantao and Qian, Jiansheng},
journal={IEEE Transactions on Multimedia},
title={Multimodal Sentiment Analysis With Image-Text Interaction Network},
year={2023},
volume={25},
pages={3375-3385},
doi={10.1109/TMM.2022.3160060}}