/HIMT

The code for our paper Hierarchical Interactive Multimodal Transformer for Aspect-Based Multimodal Sentiment Analysis

Primary LanguagePython

HIMT

The code for our TAFFC 2022 paper (https://ieeexplore.ieee.org/abstract/document/9765342/):

Jianfei Yu, Kai Chen, and Rui Xia. "Hierarchical Interactive Multimodal Transformer for Aspect-Based Multimodal Sentiment Analysis." IEEE Transactions on Affective Computing (2022).

----------Update------------

2023.03.23 We upload mywordbag.txt and myembedding.txt to Baidu Netdist with code kbtf which are needed when running our code.

Usage

We use Faster R-CNN to extract image features and corresponding semantic labels. You can download from Baidu Netdist with the extraction code kbtf. Before you run the bash file, you need to change the parameter --img_path to your saved path.

For the textual input of each sample, please refer to TomBERT to download them from the absa_data folder, and then change the parameter --data_dir to your saved path.

Training and Inference

bash run_himt.sh

Acknowledgements