LianaWang / itn

ITN Project codes and models for CVPR 2018 paper: Geometry-Aware Scene Text Detection with Instance Transformation Network
40 stars 13 forks source link

文章中的fusion层和代码中的fusion层实现不同? #5

Open BaofengZan opened 5 years ago

BaofengZan commented 5 years ago

作者你好,我看文章中的网络结构是三个feature map 送入fusion层,而在Itn-vgg中只是把conv5_3送入fusion中,这两者有区别么?还是我没有理解清楚,忘指教。谢谢。