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

2026-03-05

AI Summary

This work addresses the limitations of existing Class Activation Mapping (CAM) methods, which often suffer from noise interference, incomplete coverage, or excessive smoothing, thereby struggling to balance discriminativeness and completeness in visualizing neural network decisions. To overcome these issues, the authors propose a unified framework that integrates gradient- and region-based paradigms by synergistically combining Grad-CAM and Score-CAM. The approach employs gradient map denoising, region contribution weighting, and an adaptive pixel-level similarity fusion mechanism to generate input-adaptive, high-quality explanations. Evaluated on standard benchmarks, the method significantly outperforms current CAM techniques, achieving superior performance in both qualitative visualizations and quantitative evaluation metrics.

Technology Category

• Computer Vision: Interpretability, Explainability, and Transparency
• Machine Learning: Calibration & Uncertainty Quantification
• Constraint Satisfaction and Optimization: Satisfiability Modulo Theories

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

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.

Problem

• Class Activation Map
• Explainable AI
• Visual Explanation
• Gradient-based CAM
• Region-based CAM

Innovation

• Fusion-CAM
• Class Activation Map
• Explainable AI
• Gradient-based Visualization
• Adaptive Fusion