A global reweighting approach for cross-domain semantic segmentation

Yuhang Zhang, Shishun Tian, Muxin Liao, Guoguang Hua, Wenbin Zou, Chen Xu

Abstract: Highlights•The adaptation difficulty is different for different samples.•The unrelated sample leads to a negative transfer.•Frèchet Inception Distance shows a strong domain discrepancy capturing ability.•The proposed global reweighting strategy alleviates the negative transfer.