研究人员推出了一种新颖的注视点估计方法GMGaze，该方法利用多尺度Transformer架构并结合上下文感知条件。该方法通过早期融合图像特征和采用混合专家（MoE）设计以实现高效计算扩展，解决了现有模型的局限性。GMGaze在多个基准测试中展现了最先进的性能，在域内和跨域注视点估计任务中均提高了准确性。 AI
影响 引入了一种新的注视点估计架构，有望提高需要眼动追踪的应用的准确性和效率。
排序理由 介绍新模型架构和基准测试结果的学术论文。
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
arXiv:2605.00799v1 Announce Type: new Abstract: Gaze estimation methods commonly use facial appearances to predict the direction of a person gaze. However, previous studies show three major challenges with convolutional neural network (CNN)-based, transformer-based, and contrasti…
Gaze estimation methods commonly use facial appearances to predict the direction of a person gaze. However, previous studies show three major challenges with convolutional neural network (CNN)-based, transformer-based, and contrastive language-image pre-training (CLIP)-based meth…