Partition Explainer Next Generation in Single Image Dehazing
Akshay Juneja, Vijay Kumar, Sunil Kumar Singla · 2024
This paper investigates the field of Explainable Artificial Intelligence (xAI), focusing two well-known models, namely LIME and SHAP. It presents a methodology to refine the interpretability and transparency of image classification tasks. This research thoroughly examines feature classes in both hazy and haze-free images by utilizing the pre-built ImageNet network for comprehensive analysis using a partition explainer. A thorough examination is performed to analyze the image classification, demonstrating the complicated systems behind decision-making. In addition to that, it plays a crucial role in the identification and resolution of issues in single-image dehazing models created using machine and deep learning methods. The results of this study make a substantial contribution to the recent trend on the interpretability of models, discovering new opportunities for advancements in AI-powered image processing. The partition explainer demonstrate transparency, which has the potential to enhance and optimize image classification models. It helps to build trust and understanding between humans and machines.