FAR-AMTN: Attention Multi-Task Network for Face Attribute Recognition

2025-06-01

Computer Vision and Image Understanding

AI Summary

This work addresses the limitations of multi-task facial attribute recognition, which often suffers from a rapid increase in model parameters with the number of tasks and insufficient interaction among high-level features. To mitigate these issues, the authors propose a weight-shared group-specific attention module (WSGSA), a cross-group feature fusion mechanism (CGFF), and a dynamic weighting strategy (DWS). These components jointly reduce model complexity while enhancing semantic inter-task relationships. Experimental results demonstrate that the proposed approach achieves higher accuracy with fewer parameters on both the CelebA and LFWA datasets, outperforming current state-of-the-art methods.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Problem

• Face Attribute Recognition
• Multi-Task Networks
• Feature Interaction
• Model Generalization
• Semantic Relations

Innovation

• Attention Multi-Task Network
• Weight-Shared Group-Specific Attention
• Cross-Group Feature Fusion
• Dynamic Weighting Strategy
• Face Attribute Recognition