Rice, a vital crop for global food security, is highly susceptible to diseases, making early detection crucial for minimizing yield losses. Deep learning models require sufficient training data to avoid overfitting and ensure effective performance. This study enhances rice leaf disease detection by exploring CMR-based augmentation techniques, including CutMix, MixUp, and RandAugment. These augmentation techniques are essential as they expand the training dataset, improve model generalization, and prevent overfitting, particularly when data is limited. Although less explored in rice leaf disease detection, these techniques significantly improve data diversity and model robustness. By experimenting with various combinations of CutMix, MixUp, RandAugment, and their combinations with DenseNet201, the study demonstrates how these strategies enhance model performance. The CMR-based approach achieved an accuracy of 98%, emphasizing the effectiveness of combining advanced augmentation techniques with DenseNet201 for robust and accurate rice disease detection.