预测的时候模型test annotation

dongwu92 / AutoPortraitMatting

Tensorflow implementation of Automatic Portrait Matting on paper "Automatic Portrait Segmentation for Image Stylization"

Apache License 2.0

478 stars 151 forks source link

Open aloyschen opened 6 years ago

aloyschen commented 6 years ago

一定要给test annotation么，如果预测的时候只给image，需要改网络么，我现在预测结果都是0

Senyao-Han commented 6 years ago

不需要给test annotation, feed_dict中删掉即可