Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Gaze Embeddings for Zero-Shot Image Classification
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use GFS instead of GH or GFG gaze embeddings | Gaze-based zero-shot classification · CUB-VW | accuracy | Helped |
| Combine participant embeddings by early fusion instead of averaging | GFS gaze embeddings · CUB-VW | accuracy: 56.6% -> 69.8% | Helped |
| Combine participant embeddings by late fusion instead of averaging | GFS gaze embeddings · CUB-VW | accuracy: 56.6% -> 69.6% | Helped |
| Add gaze duration (d) to location features (ablation) | GFS gaze features · CUB-VW | accuracy | Helped |
| Add sequence angles (α1, α2) to location and duration features (ablation) | GFS gaze features · CUB-VW | accuracy | Helped |
| Add pupil diameter (R) (ablation) | GFS gaze features · CUB-VW | accuracy | No clear effect |
| Restrict the full gaze data to the images annotated with bubbles (ablation) | GFS EARLY gaze embeddings · CUB-VW | accuracy: 73.9% -> 69.7% | Hurt |
| Restrict gaze points to locations inside bubbles (ablation) | GFS EARLY gaze embeddings · CUB-VW | accuracy: 73.9% -> 64.0% | Hurt |
| Average rather than concatenate gaze points inside bubbles (ablation) | GFS EARLY gaze embeddings · CUB-VW | accuracy: 73.9% -> 55.0% | Hurt |
| Use one random gaze point inside each bubble (ablation) | GFS EARLY gaze embeddings · CUB-VW | accuracy: 73.9% -> 49.2% | Hurt |
| Combine gaze embeddings with expert attributes | Zero-shot image classification · CUB-VW | accuracy: 73.9% -> 78.2% | Helped |
| Use gaze rather than expert attributes | Zero-shot image classification · CUB-VWSW | accuracy: 42.7 -> 26.0 | Hurt |
Closest papers by meaning
- Goal-Oriented Gaze Estimation for Zero-Shot Learning (CVPR 2021 · similarity 0.90)
- Unsupervised Representation Learning for Gaze Estimation (CVPR 2020 · similarity 0.87)
- Improved Zero-Shot Classification by Adapting VLMs with Text Descriptions (CVPR 2024 · similarity 0.86)
- Gaze from Origin: Learning for Generalized Gaze Estimation by Embedding the Gaze Frontalization Process (AAAI 2024 · similarity 0.86)
- Learning from Observer Gaze: Zero-Shot Attention Prediction Oriented by Human-Object Interaction Recognition (CVPR 2024 · similarity 0.85)
- Class Semantic Attribute Perception Guided Zero-Shot Learning (AAAI 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.