SS-CAM: Smoothed Score-CAM for Sharper Visual Feature Localization

TL;DR

This paper introduces SS-CAM, an improved visualization technique for CNNs that enhances the sharpness and localization of object features, outperforming previous methods like Score-CAM in accuracy and faithfulness.

Contribution

The paper proposes SS-CAM, a novel extension of Score-CAM, incorporating smoothing to produce sharper and more accurate visual explanations of CNN decisions.

Findings

SS-CAM outperforms Score-CAM in localization accuracy.

SS-CAM provides more centralized and sharper feature visualizations.

The method improves faithfulness of explanations on the ILSVRC dataset.

Abstract

Interpretation of the underlying mechanisms of Deep Convolutional Neural Networks has become an important aspect of research in the field of deep learning due to their applications in high-risk environments. To explain these black-box architectures there have been many methods applied so the internal decisions can be analyzed and understood. In this paper, built on the top of Score-CAM, we introduce an enhanced visual explanation in terms of visual sharpness called SS-CAM, which produces centralized localization of object features within an image through a smooth operation. We evaluate our method on the ILSVRC 2012 Validation dataset, which outperforms Score-CAM on both faithfulness and localization tasks.