Saved in:
| Main Authors: | , |
|---|---|
| Format: | Recurso digital |
| Language: | |
| Published: |
Zenodo
2025
|
| Online Access: | https://doi.org/10.5281/zenodo.17819966 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Table of Contents:
- The field of computer vision has made significant strides in recent years, yet challenges remain in achieving human-level understanding and explainability, especially when reasoning about complex scenes. Deep learning models often excel at feature extraction but struggle with relational reasoning, contextual interpretation, and providing transparent decision-making processes. To address these limitations, we propose a novel approach: Neuro-Symbolic Scene Graph Reasoning for Explainable Vision. This framework integrates the perceptual strengths of neural networks with the reasoning capabilities of symbolic methods, leveraging scene graphs as a structured representation of visual information. By combining these paradigms, we aim to create a system that can not only accurately interpret visual scenes but also provide clear and logical explanations for its inferences. Our approach involves generating scene graphs from visual input using deep learning models, and then employing symbolic reasoning techniques to answer queries, make predictions, and provide justifications based on the relationships and attributes captured in the graph. We explore different neuro-symbolic architectures and reasoning algorithms to optimize performance and explainability. We demonstrate the effectiveness of our approach through experiments on various visual reasoning tasks, showcasing its ability to provide insightful explanations and improve overall accuracy.