Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2507.11522 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866916844799000576 |
|---|---|
| author | Mehmood, Tariq Ahmad, Hamza Shakeel, Muhammad Haroon Taj, Murtaza |
| author_facet | Mehmood, Tariq Ahmad, Hamza Shakeel, Muhammad Haroon Taj, Murtaza |
| contents | EEG-based brain-computer interfaces (BCIs) have shown promise in various applications, such as motor imagery and cognitive state monitoring. However, decoding visual representations from EEG signals remains a significant challenge due to their complex and noisy nature. We thus propose a novel 5-stage framework for decoding visual representations from EEG signals: (1) an EEG encoder for concept classification, (2) cross-modal alignment of EEG and text embeddings in CLIP feature space, (3) caption refinement via re-ranking, (4) weighted interpolation of concept and caption embeddings for richer semantics, and (5) image generation using a pre-trained Stable Diffusion model. We enable context-aware EEG-to-image generation through cross-modal alignment and re-ranking. Experimental results demonstrate that our method generates high-quality images aligned with visual stimuli, outperforming SOTA approaches by 13.43% in Classification Accuracy, 15.21% in Generation Accuracy and reducing Fréchet Inception Distance by 36.61%, indicating superior semantic alignment and image quality. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2507_11522 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | CATVis: Context-Aware Thought Visualization Mehmood, Tariq Ahmad, Hamza Shakeel, Muhammad Haroon Taj, Murtaza Computer Vision and Pattern Recognition Machine Learning EEG-based brain-computer interfaces (BCIs) have shown promise in various applications, such as motor imagery and cognitive state monitoring. However, decoding visual representations from EEG signals remains a significant challenge due to their complex and noisy nature. We thus propose a novel 5-stage framework for decoding visual representations from EEG signals: (1) an EEG encoder for concept classification, (2) cross-modal alignment of EEG and text embeddings in CLIP feature space, (3) caption refinement via re-ranking, (4) weighted interpolation of concept and caption embeddings for richer semantics, and (5) image generation using a pre-trained Stable Diffusion model. We enable context-aware EEG-to-image generation through cross-modal alignment and re-ranking. Experimental results demonstrate that our method generates high-quality images aligned with visual stimuli, outperforming SOTA approaches by 13.43% in Classification Accuracy, 15.21% in Generation Accuracy and reducing Fréchet Inception Distance by 36.61%, indicating superior semantic alignment and image quality. |
| title | CATVis: Context-Aware Thought Visualization |
| topic | Computer Vision and Pattern Recognition Machine Learning |
| url | https://arxiv.org/abs/2507.11522 |