Gaze Annotation

Gaze annotation involves using eye-tracking data to label images or videos, offering a less labor-intensive alternative to traditional manual annotation for training computer vision models. Current research focuses on developing methods to effectively utilize gaze data, including leveraging gaze heatmaps for weak supervision in medical image segmentation and employing generative adversarial networks (GANs) to augment existing datasets. This approach is proving valuable in various applications, from improving the efficiency of medical image analysis and gaze estimation to enhancing human motion forecasting and understanding children's developmental behaviors.