Dental caries detection using radiographs remains a critical challenge in modern dentistry due to the high variability in image quality and diagnostic subjectivity. This paper presents Color-CAM, a novel color-coded explainable AI (XAI) framework that enhances caries segmentation using an ensemble of deep learning models. Our approach employs seven distinct deep learning architectures, each generating individual segmentation masks, which are then aggregated into a color-coded confidence map. The proposed system visualizes caries probability using an intuitive multi-color scheme, where red signifies high confidence in caries presence, while white indicates no detected caries. This aggregated mask is overlaid on the original X-ray, providing dentists with an interpretable, transparent, and reliable diagnostic aid. By leveraging ensemble learning and confidence visualization, Color-CAM reduces model bias, improves segmentation accuracy, and enhances trust in AI assisted diagnostics. Experimental results demonstrate the effectiveness of our approach in providing clinically meaningful insights, bridging the gap between AI-driven automation and human decision-making in dental healthcare.