GiPL: Generative augmented iterative Pseudo-Labeling for Cross-Domain Few-Shot Object Detection

ArXi:2605.29539v1 Announce Type: cross Vision-language foundation models have shown promising zero-shot generalization for Cross-Domain Few-Shot Object Detection (CD-FSOD). However, they face two critical challenges in fine-tuning: insufficient set utilization due to sparse single-instance annotations, and severe overfitting under extremely limited target-domain samples. To address these issues, this paper proposes GiPL, an efficient two-branch