AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Facial Dynamics Interpreter Network: What are the Important Relations between Local Dynamics for Facial Trait Estimation?

Seong Tae Kim, Yong Man Ro

ECCV 2018 · Label-efficient and robust learning › Recognition and long-tail · approach: Uncertainty-aware facial expression recognition · 6 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add locational features (ablation)Facial dynamics interpreter network · UvA-NEMO age estimationMAE (years): 4.06 -> 4.00Improved
add relational importance on top of locational features (ablation)Facial dynamics interpreter network · UvA-NEMO age estimationMAE (years): 4.00 -> 3.87Improved
use relational features of pairs instead of regional-importance aggregation (ablation)Facial dynamics interpreter network · UvA-NEMO age estimationMAE (years): 4.26 -> 4.06Improved
add locational features (ablation)Facial dynamics interpreter network · UvA-NEMO gender classificationaccuracy (%): 88.79 -> 89.35Improved
add relational importance on top of locational features (ablation)Facial dynamics interpreter network · UvA-NEMO gender classificationaccuracy (%): 89.35 -> 90.08Improved
use relational features of pairs instead of regional-importance aggregation (ablation)Facial dynamics interpreter network · UvA-NEMO gender classificationaccuracy (%): 88.87 -> 88.79No clear effect
model local-region relations instead of a holistic LSTMProposed method · UvA-NEMO age estimationMAE (years): 4.02 -> 3.87Improved
model local-region relations instead of a holistic LSTMProposed method · UvA-NEMO gender classification (all)accuracy (%): 87.10 -> 90.08Improved
add facial dynamics to image-based CNNProposed method vs Image-based CNN · UvA-NEMO gender classification (all)accuracy (%): 86.94 -> 90.08Improved
perturb the most important region (right cheek) with a zero vector rather than less important parts (ablation)Facial dynamics interpreter network · UvA-NEMO age group 37-66MAE (years): 7.56 -> 9.00Hurt
perturb local dynamic features with zero vector (ablation)Facial dynamics interpreter network · UvA-NEMO age group 37-66MAE (years): 5.05 -> 7.56Hurt
use relational importance for robustness to feature perturbation of left cheek (ablation)Facial dynamics interpreter network · single 17-year-old sequenceabsolute error (years): 7.45 -> 2.38Improved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy