What question did this study set out to answer?

To enhance facial expression recognition by separating shared expression similarities from expression-specific variations using a dual-network decomposition and reconstruction framework.

June 1, 2021

Feature Decomposition and Reconstruction Learning for Effective Facial Expression Recognition

View Full Paper

Ask AI

Bookmark

Authors

DRDelian RuanXiamen University YYYan YanBeijing Institute of Fashion Technology SLShenqi LaiZhejiang University

Discussion

Member takes

Implication

Experimental evaluation reveals improved facial expression recognition accuracy across laboratory and real-world datasets, highlighting the value of feature decomposition and reconstruction.

Key Points

To enhance facial expression recognition by separating shared expression similarities from expression-specific variations using a dual-network decomposition and reconstruction framework.
Designed a Feature Decomposition Network to extract action-aware latent features representing shared expression traits from backbone representations.
Implemented a Feature Reconstruction Network with intra-feature and inter-feature relation modules to capture unique variations and reconstruct the final expression feature.
Evaluated recognition accuracy across three controlled laboratory datasets (CK+, MMI, Oulu-CASIA) and two real-world datasets (RAF-DB, SFEW).
Demonstrated consistently higher recognition accuracy than multiple state-of-the-art methods across all tested laboratory and in-the-wild image databases.
Confirmed that isolating shared expression properties from unique variations improves the discriminative power of reconstructed facial representations.

Cite This Study

Ruan et al. (2021) studied this question.

synapsesocial.com/papers/6a156b1ea2f71238514e6897 https://doi.org/10.1109/cvpr46437.2021.00757

View Full Paper

Ask AI

Bookmark

Implication

Experimental evaluation reveals improved facial expression recognition accuracy across laboratory and real-world datasets, highlighting the value of feature decomposition and reconstruction.

Feature Decomposition and Reconstruction Learning for Effective Facial Expression Recognition

Discussion

Member takes

Implication

Key Points

Cite This Study

Feature Decomposition and Reconstruction Learning for Effective Facial Expression Recognition

Discussion

Member takes

Implication

Key Points

Cite This Study

Also Consider

Also Consider