A disagreement-aware DFER framework that trains directly on the raw annotator count vector using a Dirichlet-Multinomial likelihood that provides scale-sensitive supervision for the Dirichlet concentration while preserving the predictive mean.
Abstract
Dynamic facial expression recognition (DFER) benchmarks such as DFEW provide multiple annotator votes per clip, yet most models collapse them to a majority label and cannot represent human disagreement at inference time. We propose a disagreement-aware DFER framework that trains directly on the raw annotator count vector using a Dirichlet-Multinomial likelihood. Unlike mean-only soft-label objectives, the proposed likelihood provides scale-sensitive supervision for the Dirichlet concentration while preserving the predictive mean. A separate ambiguity head predicts annotation entropy for unseen clips, and a monotone Chow-style reject rule combines predicted ambiguity, vacuity, temporal instability, and input quality for selective prediction. On DFEW, the method preserves recognition accuracy while reducing ECE by 30% and AURC by 15%, and predicted ambiguity reaches a Spearman correlation of 0.52 with the annotation entropy of test clips. The calibration and selective-prediction gains transfer to FERV39k and remain under identity- and movie-disjoint DFEW splits.
A stability-oriented convolutional study that rethinks model capacity for seven-class small-sample FER and combines stratified partitioning, grayscale normalization, compact VGG-style representation learning, global average pooling, label smoothing, dropout, L2 regularization, and momentum-based optimization to control...
Yang-Xuan Xie· Applied and Computational En...· 0 citations
Noisy annotations and class imbalance commonly coexist in real-world facial expression recognition (FER), making hard but correctly labeled minority-class samples difficult to distinguish from unreliable training samples. Existing noisy FER approaches mainly improve robustness through sample selection or re-weighting...
Facial emotion recognition (FER) is a critical component in human-computer interaction, but existing systems often suffer from limited demographic awareness and weak interpretability, as most methods overlook the influence of gender on facial expressions and rely on image-level classification without spatial localizati...
Md Sadman Haque, Md. Shakhawat Hossain, Zobaer Ibn Razzaque et al.· Discover Networks· 0 citations
A novel bias-mitigation method that decouples and regularizes age- and emotion-related components within the self-attention mechanism of Transformer to reduce age-related bias and enhance age-invariant emotion separation is proposed.
Jaeil Park, Sung-Bae Cho· International Journal of Int...· 0 citations
Control comparisons and ablations indicate that the retained model has the most favorable observed cleanness–robustness trade-off among the tested epoch-matched alternatives; however, fixed-checkpoint comparisons on Occlusion-RAF-DB are not significant after Holm correction, while broader cross-domain validation remain...
Xue-Feng Zhao, Yi-Xuan Dong, Zhao-Man Zhong et al.· Italian National Conference...· 0 citations
: Facial emotion recognition (FER) remains difficult in real-world settings. Inter-subject variability, lighting changes, occlusion, and class imbalance all limit performance. Most FER systems rely on one convolutional or transformer backbone. This narrows the features available for classification. This paper presents...
Rashid Jahangir, Nazik Alturki, M. Alreshoodi· Computer Modeling in Enginee...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
