Explainable Object Detection in 360° Images Through Fuzzy Logic Systems and Vision-Language Models Integration
Amer Harfoush, Hani Hagras, Hugo Leon-Garza, Anasol Peña-Ríos
University of Essex BT Group (United Kingdom)
阅读操作
确认中在文库中上传 PDF 后可生成中文音频讲解。
摘要与影响
Panoramic images (360-degree) contain significant amount of information since they manage to take a snapshot of the complete surroundings at once which makes them a great choice in multiple fields. However, they pose special challenges for object detection due to significant geometric distortions coming from equirectangular projection. Current methods are inadequate for critical applications since they do not have the needed accuracy and provide no explanations for their decisions. This paper presents a novel framework that integrates Fuzzy Logic Systems (FLSs), an object detection model, and Vision-Language Models (VLMs) to achieve both good detection performance and explainability. Our method uses a fine-tuned YOLOv9 object detector enhanced by a VLM-based validation engine. The fuzzy logic system module evaluates image quality and adapts processing parameters, providing linguistic explanations that bridge numerical processing with human understanding. Testing on industrial datasets from British Telecom demonstrates robust performance on previously unseen object categories, validating real-world applicability. The proposed framework achieves up to 124.67% improvement over previous state-of-the-art methods while providing explanations for each detection decision.
逐年被引趋势
暂无年度引用数据
关键指标
同类平均 = 1
同领域 · 同年份 · 同类型
Google Scholar 与 OpenAlex 的被引统计范围不同,数值存在差异属正常。
AI 辅助阅读
依据:摘要
可就本文提问;依据不足时会说明。
学术脉络
学科主题
计算机 / AIAdvanced Neural Network Applications
Explainable Artificial Intelligence (XAI) · Multimodal Machine Learning Applications