# BHGap ยท Trained Checkpoints
### A Deep Iterative Prompting and Multi-stage Alignment Framework for Dynamic Facial Expression Recognition
> ๐ค This repository hosts the **trained model weights** for **BHGap** (WWW 2026 Oral).
> For the full code, training and evaluation pipeline, please visit the **[GitHub repository](https://github.com/NDYZD666/-public-BHGap)**.
## ๐ Abstract
Dynamic Facial Expression Recognition (DFER) is a crucial part of affective computing, with broad applications in human-computer interaction and social media content analysis. Effectively integrating audio-visual signals remains the core challenge, as existing approaches are constrained by **(1) shallow, static fusion** that fails to capture the dynamic co-evolution of features, and **(2) implicit, coarse alignment** that cannot bridge the modality gap.
**BHGap** reformulates audio-visual collaboration from a one-shot fusion event into a continuous, reciprocal generation process spanning every layer of frozen backbone encoders:
- **SDIC** โ an SSM (Mamba)-based **Cross-Modal Prompt Generator** dynamically produces *guidance prompts* for the counterpart modality at each encoding layer, enabling deep and fine-grained feature co-evolution.
- **MSC2F** โ a coarse-to-fine alignment module that combines **low-rank adversarial alignment** (macro-level distribution & spatio-temporal congruity) with **MMD-driven implicit differentiation** (micro-level statistical & semantic consistency).
It achieves **state-of-the-art** performance on **DFEW** and **MAFW** using raw audio-visual inputs only.
## ๐ Results
Evaluated under 5-fold cross-validation (**WAR** = Weighted Average Recall, **UAR** = Unweighted Average Recall).
| Dataset | Modality | WAR (%) | UAR (%) |
| --- | :---: | :---: | :---: |
| **DFEW** | Audio + Visual | **78.80** | **69.03** |
| **MAFW** | Audio + Visual | **59.97** | **47.68** |
## ๐ฆ Checkpoints
This repository provides the trained BHGap checkpoints for **DFEW** and **MAFW** (5-fold). Each fold produces the best models selected by WAR / UAR (`model_best_war.pth`, `model_best_uar.pth`); see the **Files and versions** tab for the exact contents.
## ๐ Usage
Download the checkpoints from the Hub:
```bash
huggingface-cli download NiDeYingZiD/BHGap-ckpt --local-dir ./checkpoints
```
```python
from huggingface_hub import snapshot_download
snapshot_download(repo_id="NiDeYingZiD/BHGap-ckpt", local_dir="./checkpoints")
```
Then evaluate with the code from the [GitHub repository](https://github.com/NDYZD666/-public-BHGap):
```bash
python evaluate.py --dataset DFEW --checkpoint ./checkpoints/model_best_war.pth --fold 5
```
> **Note** โ These are task checkpoints for the BHGap pipeline (frozen MAE-Face / AudioMAE backbones + trainable SDIC & MSC2F modules), not a standalone `transformers` model. Please load them through the BHGap code.
## ๐ Citation
```bibtex
@inproceedings{zhang2026bhgap,
title = {BHGap: A Deep Iterative Prompting and Multi-stage Alignment Framework for Dynamic Facial Expression Recognition},
author = {Zhang, Yichi and Han, Yunqi and Ding, Jiayue and Chen, Liangyu},
booktitle = {Proceedings of the ACM Web Conference 2026 (WWW '26)},
year = {2026},
address = {Dubai, United Arab Emirates},
publisher = {Association for Computing Machinery},
doi = {10.1145/3774904.3792417}
}
```
## ๐ Acknowledgements
Built upon [MAE-Face](https://github.com/FuxiVirtualHuman/MAE-Face), [AudioMAE](https://github.com/facebookresearch/AudioMAE), [Mamba](https://github.com/state-spaces/mamba), [MMA-DFER](https://github.com/katerynaCh/MMA-DFER), and the [DFEW](https://dfew-dataset.github.io/) / [MAFW](https://mafw-database.github.io/MAFW/) datasets. Thank you for these excellent efforts!
-1f6feb?style=flat-square){kind=icon}
