Ground Truth refers to the actual, verified values or outcomes used as a reference for training and evaluating machine learning models. High-quality ground truth is essential for accurate model assessment.