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

📅 2026-03-15
📈 Citations: 0
Influential: 0
📄 PDF
🤖 AI Summary
This study addresses the challenge of recognizing ambivalence/hesitancy (A/H) states in video-recorded individuals by proposing a multimodal fusion approach based on inter-modal embedding discrepancies. The method explicitly models cross-modal inconsistency—a hallmark of A/H—by computing absolute differences among visual, audio, and textual embeddings in a shared latent space. Visual features are derived using Py-Feat to extract action units (AUs), audio representations are obtained via Wav2Vec 2.0, and textual inputs are encoded with BERT; these modalities are subsequently aligned through a BiLSTM combined with attention-based pooling. Evaluated on the BAH dataset, the model achieves a Macro F1 score of 0.6808, substantially outperforming baseline methods (0.2827). Further analysis confirms that temporal variability in AUs serves as a critical visual discriminative cue for A/H detection.

Technology Category

Application Category

📝 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.
Problem

Research questions and friction points this paper is trying to address.

Ambivalence
Hesitancy
Video Recognition
Multimodal Fusion
Affective Behavior Analysis
Innovation

Methods, ideas, or system contributions that make the work stand out.

divergence-based fusion
multimodal conflict
Action Units
BiLSTM with attention
ambivalence/hesitancy recognition
💼 Related Jobs
No related jobs found.
A
Aislan Gabriel O. Souza
Universidade de Pernambuco (UPE), Escola Politécnica de Pernambuco (POLI), Recife, Brazil
A
Agostinho Freire
Universidade de Pernambuco (UPE), Escola Politécnica de Pernambuco (POLI), Recife, Brazil
L
Leandro Honorato Silva
Universidade de Pernambuco (UPE), Escola Politécnica de Pernambuco (POLI), Recife, Brazil
I
Igor Lucas B. da Silva
Universidade de Pernambuco (UPE), Escola Politécnica de Pernambuco (POLI), Recife, Brazil
J
João Vinícius R. de Andrade
Universidade de Pernambuco (UPE), Escola Politécnica de Pernambuco (POLI), Recife, Brazil
G
Gabriel C. de Albuquerque
Universidade de Pernambuco (UPE), Escola Politécnica de Pernambuco (POLI), Recife, Brazil
L
Lucas Matheus da S. Oliveira
Universidade de Pernambuco (UPE), Escola Politécnica de Pernambuco (POLI), Recife, Brazil
M
Mário Stela Guerra
Universidade de Pernambuco (UPE), Escola Politécnica de Pernambuco (POLI), Recife, Brazil
L
Luciana Machado
Universidade de Pernambuco (UPE), Escola Politécnica de Pernambuco (POLI), Recife, Brazil