Landmark-Guided Gaze Estimation via Conditional Keypoint Generation and Cross-Attention Fusion
Open Access
- 25 February 2026
- journal article
- research article
- Published by MDPI AG in Information
- Vol. 17 (3) , 224
Abstract
In gaze estimation, existing mainstream methods face significant challenges in capturing the fine-grained structures of eye regions, particularly in the absence of explicit geometric prior information, which hampers gaze prediction accuracy. To address this limitation, we propose the landmark-guided gaze estimation network (LGNet), a gaze estimation method guided by keypoints, which effectively incorporates geometric prior information to enhance estimation performance. The proposed method begins by training an eye-keypoint generator on the synthetic UnityEyes dataset using a Conditional Variational Autoencoder (CVAE). Next, we introduce a Symmetric Spatial Feature Fusion module (SSFF), combined with a dual-stream cross-attention mechanism, to achieve semantic alignment between the keypoint features and the facial image features extracted using ResNet50. Furthermore, we propose a Gated Channel Reweighting module (GCR) to suppress redundant information and amplify the critical features, thereby enhancing the model’s overall response. Experimental results demonstrate that LGNet outperforms existing methods on three benchmark datasets. The code for this research has been made publicly available.Keywords
Funding Information
- National Natural Science Foundation of China (62507035)
- National Natural Science Foundation of China (62307034)
This publication has 23 references indexed in Scilit:
- Multi-task driver gaze estimation in real world driving scenesEngineering Applications of Artificial Intelligence, 2025
- ADGaze: Anisotropic Gaussian Label Distribution Learning for fine-grained gaze estimationPattern Recognition, 2025
- Cascaded learning with transformer for simultaneous eye landmark, eye state and gaze estimationPattern Recognition, 2024
- DADL: Double Asymmetric Distribution Learning for head pose estimation in wisdom museumJournal of King Saud University Computer and Information Sciences, 2024
- Gaze Estimation by Attention-Induced Hierarchical Variational Auto-EncoderIEEE Transactions on Cybernetics, 2023
- Attention-guided and fine-grained feature extraction from face images for gaze estimationEngineering Applications of Artificial Intelligence, 2023
- GCANet: Geometry cues-aware facial expression recognition based on graph convolutional networksJournal of King Saud University Computer and Information Sciences, 2023
- MFDNet: Collaborative Poses Perception and Matrix Fisher Distribution for Head Pose EstimationIEEE Transactions on Multimedia, 2021
- Deep Pictorial Gaze Estimation2018
- MPIIGaze: Real-World Dataset and Deep Appearance-Based Gaze EstimationIEEE Transactions on Pattern Analysis and Machine Intelligence, 2017