Reducing false positives in strong lens detection through effective augmentation and ensemble learning