Learning Transferable Facial Emotion Representations from Large-Scale Semantically Rich Captions

TL;DR

This paper introduces EmoCap100K, a large-scale dataset with rich emotional captions, and EmoCapCLIP, a contrastive learning framework that leverages these captions to improve facial emotion recognition across multiple benchmarks.

Contribution

The paper presents a novel large-scale dataset and a contrastive learning method that effectively utilize semantically rich captions for facial emotion representation learning.

Findings

Superior performance on 20+ benchmarks

Effective exploitation of multi-level caption information

Enhanced generalization in emotion recognition

Abstract

Current facial emotion recognition systems are predominately trained to predict a fixed set of predefined categories or abstract dimensional values. This constrained form of supervision hinders generalization and applicability, as it reduces the rich and nuanced spectrum of emotions into oversimplified labels or scales. In contrast, natural language provides a more flexible, expressive, and interpretable way to represent emotions, offering a much broader source of supervision. Yet, leveraging semantically rich natural language captions as supervisory signals for facial emotion representation learning remains relatively underexplored, primarily due to two key challenges: 1) the lack of large-scale caption datasets with rich emotional semantics, and 2) the absence of effective frameworks tailored to harness such rich supervision. To this end, we introduce EmoCap100K, a large-scale facial emotion caption dataset comprising over 100,000 samples, featuring rich and structured semantic descriptions that capture both global affective states and fine-grained local facial behaviors. Building upon this dataset, we further propose EmoCapCLIP, which incorporates a joint global-local contrastive learning framework enhanced by a cross-modal guided positive mining module. This design facilitates the comprehensive exploitation of multi-level caption information while accommodating semantic similarities between closely related expressions. Extensive evaluations on over 20 benchmarks covering five tasks demonstrate the superior performance of our method, highlighting the promise of learning facial emotion representations from large-scale semantically rich captions. The code and data will be available at https://github.com/sunlicai/EmoCapCLIP.