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 7 特征增强模块对模型性能的影响

Tab.7 Impact of feature enhancement module on model performance

特征增强 图像编码器 $N_{\mathrm{p}} / 10^6 $ S4 MS3 $ {M_{\mathrm{J}}} $/% $ {M_{\mathrm{F}}} $/% $ {M_{\mathrm{J}}} $/% $ {M_{\mathrm{F}}} $/% 无 ResNet-50 120.75 77.59 87.5 53.37 64.0 PVT v2 177.98 84.24 91.4 59.96 71.6 有 ResNet-50 136.45 78.68 88.0 59.50 69.8 PVT v2 179.72 84.43 91.6 65.16 75.6