Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Fusion-CAM: Integrating Gradient and Region-Based Class Activation Maps for Robust Visual Explanations

Hajar Dekdegue, Moncef Garouani, Josiane Mothe, Jordan Bernigaud

TL;DR

Fusion-CAM is introduced, a novel framework that bridges this explanatory gap by unifying both paradigms through a dedicated fusion mechanism to produce robust and highly discriminative visual explanations, providing a robust and flexible tool for interpreting deep neural networks.

Abstract

Interpreting the decision-making process of deep convolutional neural networks remains a central challenge in achieving trustworthy and transparent artificial intelligence. Explainable AI (XAI) techniques, particularly Class Activation Map (CAM) methods, are widely adopted to visualize the input regions influencing model predictions. Gradient-based approaches (e.g. Grad-CAM) provide highly discriminative, fine-grained details by computing gradients of class activations but often yield noisy and incomplete maps that emphasize only the most salient regions rather than the complete objects. Region-based approaches (e.g. Score-CAM) aggregate information over larger areas, capturing broader object coverage at the cost of over-smoothing and reduced sensitivity to subtle features. We introduce Fusion-CAM, a novel framework that bridges this explanatory gap by unifying both paradigms through a dedicated fusion mechanism to produce robust and highly discriminative visual explanations. Our method first denoises gradient-based maps, yielding cleaner and more focused activations. It then combines the refined gradient map with region-based maps using contribution weights to enhance class coverage. Finally, we propose an adaptive similarity-based pixel-level fusion that evaluates the agreement between both paradigms and dynamically adjusts the fusion strength. This adaptive mechanism reinforces consistent activations while softly blending conflicting regions, resulting in richer, context-aware, and input-adaptive visual explanations. Extensive experiments on standard benchmarks show that Fusion-CAM consistently outperforms existing CAM variants in both qualitative visualization and quantitative evaluation, providing a robust and flexible tool for interpreting deep neural networks.

Fusion-CAM: Integrating Gradient and Region-Based Class Activation Maps for Robust Visual Explanations

TL;DR

Fusion-CAM is introduced, a novel framework that bridges this explanatory gap by unifying both paradigms through a dedicated fusion mechanism to produce robust and highly discriminative visual explanations, providing a robust and flexible tool for interpreting deep neural networks.

Abstract

Interpreting the decision-making process of deep convolutional neural networks remains a central challenge in achieving trustworthy and transparent artificial intelligence. Explainable AI (XAI) techniques, particularly Class Activation Map (CAM) methods, are widely adopted to visualize the input regions influencing model predictions. Gradient-based approaches (e.g. Grad-CAM) provide highly discriminative, fine-grained details by computing gradients of class activations but often yield noisy and incomplete maps that emphasize only the most salient regions rather than the complete objects. Region-based approaches (e.g. Score-CAM) aggregate information over larger areas, capturing broader object coverage at the cost of over-smoothing and reduced sensitivity to subtle features. We introduce Fusion-CAM, a novel framework that bridges this explanatory gap by unifying both paradigms through a dedicated fusion mechanism to produce robust and highly discriminative visual explanations. Our method first denoises gradient-based maps, yielding cleaner and more focused activations. It then combines the refined gradient map with region-based maps using contribution weights to enhance class coverage. Finally, we propose an adaptive similarity-based pixel-level fusion that evaluates the agreement between both paradigms and dynamically adjusts the fusion strength. This adaptive mechanism reinforces consistent activations while softly blending conflicting regions, resulting in richer, context-aware, and input-adaptive visual explanations. Extensive experiments on standard benchmarks show that Fusion-CAM consistently outperforms existing CAM variants in both qualitative visualization and quantitative evaluation, providing a robust and flexible tool for interpreting deep neural networks.
Paper Structure (23 sections, 6 equations, 5 figures, 11 tables, 1 algorithm)

This paper contains 23 sections, 6 equations, 5 figures, 11 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Fusion-CAM approach
  4. Gradient-based CAM Denoising
  5. Combination of Denoised Gradient and Region-based CAM
  6. Similarity Based Fusion
  7. Experimental Study
  8. Datasets
  9. Base Models and Activation Maps
  10. Baselines and Evaluation Metrics
  11. Results and Discussion
  12. Fusion-Cam is effective
  13. Sensitivity - Threshold Parameter $\theta$
  14. Efficiency
  15. Ablation study
  16. ...and 8 more sections

Figures (5)

  • Figure 1: Overview of Fusion-CAM.
  • Figure 2: Comparison of different CAM methods on examples (One instance for the two first rows, several instances for the last).
  • Figure 3: Visualization of insertion and deletion curves across multiple XAI techniques.
  • Figure 4: Additional qualitative visualizations comparing FusionCAM with existing methods across different object categories and domains. FusionCAM produces more spatially comprehensive and detailed activation maps, effectively capturing fine structural and contextual features.
  • Figure 5: Insertion and deletion curves for the ensemble CAM techniques Union-CAM and Fusion-CAM.