mirror of
https://github.com/open-mmlab/mmsegmentation.git
synced 2025-06-03 01:05:51 +08:00
dd47cef801
## Motivation Support SOTA real-time semantic segmentation method in [Paper with code](https://paperswithcode.com/task/real-time-semantic-segmentation) Paper: https://arxiv.org/pdf/2206.02066.pdf Official repo: https://github.com/XuJiacong/PIDNet ## Current results **Cityscapes** |Model|Ref mIoU|mIoU (ours)| |---|---|---| |PIDNet-S|78.8|78.74| |PIDNet-M|79.9|80.22| |PIDNet-L|80.9|80.89| ## TODO - [x] Support inference with official weights - [x] Support training on Cityscapes - [x] Update docstring - [x] Add unit test
11 lines
434 B
Python
11 lines
434 B
Python
_base_ = './pidnet-s_2xb6-120k_1024x1024-cityscapes.py'
|
|
checkpoint_file = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/pidnet/pidnet-l_imagenet1k_20230306-67889109.pth' # noqa
|
|
model = dict(
|
|
backbone=dict(
|
|
channels=64,
|
|
ppm_channels=112,
|
|
num_stem_blocks=3,
|
|
num_branch_blocks=4,
|
|
init_cfg=dict(checkpoint=checkpoint_file)),
|
|
decode_head=dict(in_channels=256, channels=256))
|