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.

Current Facial expression recognition methods typically extract facial features indiscriminately, incorporating expression-irrelevant information that compromises recognition accuracy. To overcome this, we propose Multi-stage Feature Sparse Constraints (MFSC), a novel model that integrates a Multi-scale Attention-based Sparse Window Selection (MSAWS) mechanism with key region graph learning. Notably, MFSC operates without dependence on pre-extracted facial landmarks, enabling more flexible deployment. The MSAWS mechanism progressively filters redundant features through multi-stage sparse attention, adaptively selecting the most discriminative facial patches. The selected tokens are structured into a dynamic graph to model regional relationships via graph neural networks (GNNs). Critically, our framework further introduces a global-guided fusion module, which effectively integrates fine-grained local features from an IR50 backbone with the global topological features from the GNN through cross-attention. This integration enables complementary strengths, where local details are enhanced by global semantic context. Comprehensive experiments on RAF-DB, FER2013, and AffectNet-7 datasets demonstrate MFSC’s superior performance, achieving state-of-the-art accuracy of 92.31%, 76.21%, and 67.35%, respectively. These results validate the effectiveness of our approach in focusing computational resources on expression-salient regions while maintaining a lightweight and efficient architecture.