Research map / Label-efficient and robust learning / Semi-supervised learning
Gaussian-Based Instance-Adaptive Intensity Modeling for Point-Supervised Facial Expression Spotting
Yicheng Deng, Hideaki Hayashi, Hajime Nagahara
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes a Gaussian-based instance-adaptive intensity modeling approach for point-supervised facial expression spotting that replaces hard pseudo-labels to better separate neutral and expression frames of varying intensity.
Problem
Point-supervised facial expression spotting (P-FES) localizes expression instances in untrimmed videos using only one timestamp label per instance; sparse labels make hard pseudo-labeling confuse neutral and expression frames of varying intensity.
Approach
Models expression intensity with Gaussian-based, instance-adaptive soft labels instead of hard pseudo-labels propagated from point annotations, so that frames near a point label receive graded intensity targets adapted to each instance.
Experimental setup
Builds on: point-supervised temporal action localization
Closest papers by meaning
- Weakly Supervised Micro- and Macro-Expression Spotting Based on Multi-Level Consistency (TPAMI 2025 · similarity 0.86)
- FERV39k: A Large-Scale Multi-Scene Dataset for Facial Expression Recognition in Videos (CVPR 2022 · similarity 0.85)
- An Enhanced Adaptive Confidence Margin for Semi-Supervised Facial Expression Recognition (TPAMI 2025 · similarity 0.85)
- Learning Adaptive Spatial Coherent Correlations for Speech-Preserving Facial Expression Manipulation (CVPR 2024 · similarity 0.84)
- TEASER: Token Enhanced Spatial Modeling for Expressions Reconstruction (ICLR 2025 · similarity 0.83)
- PrefAce: Face-Centric Pretraining with Self-Structure Aware Distillation (AAAI 2024 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
