EasyCV/configs/segmentation/segformer/segformer_b5_coco.py

53 lines
1.8 KiB
Python

_base_ = './segformer_b0_coco.py'
model = dict(
pretrained=
'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segformer/mit_b5_20220624-658746d9.pth',
backbone=dict(
embed_dims=64,
num_layers=[3, 6, 40, 3],
),
decode_head=dict(
in_channels=[64, 128, 320, 512],
channels=768,
),
)
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
crop_size = (640, 640)
train_pipeline = [
dict(type='MMResize', img_scale=(2048, 640), ratio_range=(0.5, 2.0)),
dict(type='SegRandomCrop', crop_size=crop_size, cat_max_ratio=0.75),
dict(type='MMRandomFlip', flip_ratio=0.5),
dict(type='MMPhotoMetricDistortion'),
dict(type='MMNormalize', **img_norm_cfg),
dict(type='MMPad', size=crop_size, pad_val=dict(img=0, masks=0, seg=255)),
dict(type='DefaultFormatBundle'),
dict(
type='Collect',
keys=['img', 'gt_semantic_seg'],
meta_keys=('filename', 'ori_filename', 'ori_shape', 'img_shape',
'pad_shape', 'scale_factor', 'flip', 'flip_direction',
'img_norm_cfg')),
]
test_pipeline = [
dict(
type='MMMultiScaleFlipAug',
img_scale=(2048, 640),
# img_ratios=[0.5, 0.75, 1.0, 1.25, 1.5, 1.75],
flip=False,
transforms=[
dict(type='MMResize', keep_ratio=True),
dict(type='MMRandomFlip'),
dict(type='MMNormalize', **img_norm_cfg),
dict(type='ImageToTensor', keys=['img']),
dict(
type='Collect',
keys=['img'],
meta_keys=('filename', 'ori_filename', 'ori_shape',
'img_shape', 'pad_shape', 'scale_factor', 'flip',
'flip_direction', 'img_norm_cfg')),
])
]