PAGE: Towards Practical Human-level Gaze Target Estimation
2026-07-06 • Computer Vision and Pattern Recognition
Computer Vision and Pattern RecognitionHuman-Computer Interaction
AI summaryⓘ
The authors developed a new model called PaGE to predict where a person is looking in a scene, which helps understand human attention. Their model combines information about the scene and the person's head features better than before. They trained a strong main model and then made smaller versions using many unlabeled images so the smaller models can work well too. These improvements let their model perform better than humans in most tests and still run efficiently for real-world devices like robots. The authors also provide the code and models for others to use.
gaze estimationhuman attentionscene semanticshead posevision transformermodel distillationunlabeled dataspatial reasoningdeep learningcomputer vision
Authors
Zhoutong Ye, Chengwen Zhang, Zhaibin Cui, Mingze Sun, Jiaqi Liu, Xiangwu Li, Qingyang Wan, Chang Liu, Xutong Wang, Huan-ang Gao, Yu Mei, Chun Yu, Yuanchun Shi
Abstract
Gaze target estimation, the task of predicting where a person is looking in a scene, is crucial to understanding human attention and intent. It is a challenging task that combines high-level understanding of global scene semantics and precise spatial reasoning using human appearance (e.g. pose, eye orientation). As a result, human-level performance remains elusive for existing models, limiting their practical application. To this end, we propose PaGE (Practical Gaze Estimator), a gaze estimation model that explicitly models the complex interaction between scene and head features. Using a PaGE model with a large ViT-H+ backbone as the teacher, we further distill student models with lighter backbones on a much larger and more diverse unlabeled dataset. The architectural improvements and novel training recipe allow PaGE to achieve state-of-the-art performance on several gaze estimation tasks, outperforming humans in 7 out of 9 metrics while reducing the human-AI gap by at least 60% in the remaining 2. The distilled student models retain most of the teacher's performance while being lightweight enough for practical deployment on robots and consumer devices. The code and model checkpoints are available at our project page.