基于对比学习的声源定位引导视听分割模型

黄文湖,赵邢,谢亮,梁浩然,梁荣华

Contrastive learning-based sound source localization-guided audio-visual segmentation model

Wenhu HUANG,Xing ZHAO,Liang XIE,Haoran LIANG,Ronghua LIANG

表 8 对比损失和定位损失对模型性能的影响

Tab.8 Impact of contrastive loss and location loss on model performance

$ {L_{{\text{loc}}}} $ $ {L_{{\text{cts}}}} $ S4 MS3 $ {M_{\text{J}}} $/% $ {M_{\text{F}}} $/% $ {M_{\text{J}}} $/% $ {M_{\text{F}}} $/% 无 无 78.42 87.9 57.62 68.1 有 无 78.34 87.9 59.11 69.8 有 有 78.68 88.0 59.50 69.8