# Solution for 10th Competition on Ambivalence/Hesitancy (AH) Video Recognition Challenge using Divergence-Based Multimodal Fusion

Source: [Solution for 10th Competition on Ambivalence/Hesitancy (AH) Video Recognition Challenge using Divergence-Based Multimodal Fusion](<https://www.sota2.com/research/paper/69bb82e30636e5bf4afc181b>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-03-20T05:00:04.947Z

Authors: Aislan Gabriel O. Souza, Agostinho Freire, Leandro Honorato Silva, Igor Lucas B. da Silva, Jo\\~ao Vin\\'icius R. de Andrade, Gabriel C. de Albuquerque, Lucas Matheus da S. Oliveira, M\\'ario Stela Guerra, Luciana Machado

Published: 2026-03-15T15:37:25.000Z

## Abstract

We address the Ambivalence/Hesitancy (A/H) Video Recognition Challenge at the 10th ABAW Competition (CVPR 2026). We propose a divergence-based multimodal fusion that explicitly measures cross-modal conflict between visual, audio, and textual channels. Visual features are encoded as Action Units (AUs) extracted via Py-Feat, audio via Wav2Vec 2.0, and text via BERT. Each modality is processed by a BiLSTM with attention pooling and projected into a shared embedding space. The fusion module computes pairwise absolute differences between modality embeddings, directly capturing the incongruence that characterizes A/H. On the BAH dataset, our approach achieves a Macro F1 of 0.6808 on the validation test set, outperforming the challenge baseline of 0.2827. Statistical analysis across 1{,}132 videos confirms that temporal variability of AUs is the dominant visual discriminator of A/H.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Video-level A/H Recognition on BAH (dev val)](<https://www.sota2.com/research/sota/video-level-a-h-recognition-on-bah-dev-val.md>): Video-level A/H Recognition; BAH (dev val); MF1: 69.12.

- [Video-level A/H Recognition on BAH (test)](<https://www.sota2.com/research/sota/video-level-a-h-recognition-on-bah-test.md>): Video-level A/H Recognition; BAH (test); MF1: not recorded.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2603.16939>)

[PDF](<https://arxiv.org/pdf/2603.16939.pdf>)