Summary of Learning User Embeddings From Human Gaze For Personalised Saliency Prediction, by Florian Strohm et al.
Learning User Embeddings from Human Gaze for Personalised Saliency Prediction
by Florian Strohm, Mihai Bâce, Andreas Bulling
First submitted to arxiv on: 20 Mar 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary A novel method is presented for generating reusable embeddings of user behavior that can be used for personalized saliency prediction tasks. The approach uses a Siamese convolutional neural encoder to learn user embeddings by contrasting pairs of natural images and corresponding saliency maps generated from limited eye tracking data. Evaluations on two public datasets show that the generated embeddings have high discriminative power, can refine universal saliency maps for individual users, and generalize well across users and images. This work has implications for other applications that require encoding individual user characteristics. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary A new way to understand how people look at pictures is introduced. It uses a special type of AI called a Siamese neural network to create a representation of each person’s viewing habits based on limited eye tracking data. This “user embedding” can then be used to improve the accuracy of image analysis models for individual users. The method was tested on two different datasets and showed promising results, including being able to refine general-purpose image analysis models to better fit an individual’s viewing style. |
Keywords
» Artificial intelligence » Embedding » Encoder » Neural network » Tracking
