kazuto1011 / deeplab-pytorch

PyTorch re-implementation of DeepLab v2 on COCO-Stuff / PASCAL VOC datasets

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Apply multi-scale during testing

TyroneLi opened this issue · comments

Thanks, I noticed that the code only support one scale image inference, could we implement multi-scale image size for inference?

The models named with "MSC" do multi-scale inference. See libs/models/msc.py.

def DeepLabV2_ResNet101_MSC(n_classes):
return MSC(
base=DeepLabV2(
n_classes=n_classes, n_blocks=[3, 4, 23, 3], atrous_rates=[6, 12, 18, 24]
),
scales=[0.5, 0.75],
)