Wenjun Li. "Scene-aware contrastive learning with multi-scale domain-invariant feature alignment for robust visual recognition under complex scenes." Scientific Reports (2026). https://doi.org/10.1038/s41598-026-57271-1