add new model

0d97cc8c · Sugon_ldc · 0d97cc8c · 0d97cc8c · 0d97cc8c · 0d97cc8c
Commit 0d97cc8c authored Jun 07, 2023 by Sugon_ldc
20 changed files
--- a/configs/quick_start/bisenet_optic_disc_512x512_1k.yml
+++ b/configs/quick_start/bisenet_optic_disc_512x512_1k.yml
+batch_size: 4
+iters: 1000
+
+train_dataset:
+  type: OpticDiscSeg
+  dataset_root: data/optic_disc_seg
+  transforms:
+    - type: Resize
+      target_size: [512, 512]
+    - type: RandomHorizontalFlip
+    - type: Normalize
+  mode: train
+
+val_dataset:
+  type: OpticDiscSeg
+  dataset_root: data/optic_disc_seg
+  transforms:
+    - type: Normalize
+  mode: val
+
+optimizer:
+  type: sgd
+  momentum: 0.9
+  weight_decay: 4.0e-5
+
+lr_scheduler:
+  type: PolynomialDecay
+  learning_rate: 0.01
+  end_lr: 0
+  power: 0.9
+
+loss:
+  types:
+    - type: CrossEntropyLoss
+  coef: [1, 1, 1, 1, 1]
+
+model:
+  type: BiSeNetV2
+  pretrained: Null
--- a/configs/quick_start/deeplabv3p_resnet18_os8_optic_disc_512x512_1k_student.yml
+++ b/configs/quick_start/deeplabv3p_resnet18_os8_optic_disc_512x512_1k_student.yml
+batch_size: 4
+iters: 1000
+
+train_dataset:
+  type: OpticDiscSeg
+  dataset_root: data/optic_disc_seg
+  transforms:
+    - type: Resize
+      target_size: [512, 512]
+    - type: RandomHorizontalFlip
+    - type: Normalize
+  mode: train
+
+val_dataset:
+  type: OpticDiscSeg
+  dataset_root: data/optic_disc_seg
+  transforms:
+    - type: Resize
+      target_size: [512, 512]
+    - type: Normalize
+  mode: val
+
+optimizer:
+  type: sgd
+  momentum: 0.9
+  weight_decay: 4.0e-5
+
+lr_scheduler:
+  type: PolynomialDecay
+  learning_rate: 0.01
+  end_lr: 0
+  power: 0.9
+
+loss:
+  types:
+    - type: CrossEntropyLoss
+  coef: [1]
+
+# distill_loss is used for distillation
+distill_loss:
+  types:
+    - type: KLLoss
+  coef: [3]
+
+model:
+  type: DeepLabV3P
+  backbone:
+    type: ResNet18_vd
+    output_stride: 8
+    multi_grid: [1, 2, 4]
+    pretrained: Null
+  num_classes: 2
+  backbone_indices: [0, 3]
+  aspp_ratios: [1, 12, 24, 36]
+  aspp_out_channels: 256
+  align_corners: False
+  pretrained: Null
--- a/configs/quick_start/deeplabv3p_resnet50_os8_optic_disc_512x512_1k_teacher.yml
+++ b/configs/quick_start/deeplabv3p_resnet50_os8_optic_disc_512x512_1k_teacher.yml
+batch_size: 4
+iters: 1000
+
+train_dataset:
+  type: OpticDiscSeg
+  dataset_root: data/optic_disc_seg
+  transforms:
+    - type: Resize
+      target_size: [512, 512]
+    - type: RandomHorizontalFlip
+    - type: Normalize
+  mode: train
+
+val_dataset:
+  type: OpticDiscSeg
+  dataset_root: data/optic_disc_seg
+  transforms:
+    - type: Resize
+      target_size: [512, 512]
+    - type: Normalize
+  mode: val
+
+optimizer:
+  type: sgd
+  momentum: 0.9
+  weight_decay: 4.0e-5
+
+lr_scheduler:
+  type: PolynomialDecay
+  learning_rate: 0.01
+  end_lr: 0
+  power: 0.9
+
+loss:
+  types:
+    - type: CrossEntropyLoss
+  coef: [1]
+
+model:
+  type: DeepLabV3P
+  backbone:
+    type: ResNet50_vd
+    output_stride: 8
+    multi_grid: [1, 2, 4]
+    pretrained: https://bj.bcebos.com/paddleseg/dygraph/resnet50_vd_ssld_v2.tar.gz
+  num_classes: 2
+  backbone_indices: [0, 3]
+  aspp_ratios: [1, 12, 24, 36]
+  aspp_out_channels: 256
+  align_corners: False
+  pretrained: Null
--- a/configs/quick_start/pp_liteseg_optic_disc_512x512_1k.yml
+++ b/configs/quick_start/pp_liteseg_optic_disc_512x512_1k.yml
+batch_size: 4
+iters: 1000
+
+train_dataset:
+  type: Dataset
+  dataset_root: data/optic_disc_seg
+  train_path: data/optic_disc_seg/train_list.txt
+  num_classes: 2
+  mode: train
+  transforms:
+    - type: ResizeStepScaling
+      min_scale_factor: 0.5
+      max_scale_factor: 2.0
+      scale_step_size: 0.25
+    - type: RandomPaddingCrop
+      crop_size: [512, 512]
+    - type: RandomHorizontalFlip
+    - type: RandomDistort
+      brightness_range: 0.5
+      contrast_range: 0.5
+      saturation_range: 0.5
+    - type: Normalize
+
+val_dataset:
+  type: Dataset
+  dataset_root: data/optic_disc_seg
+  val_path: data/optic_disc_seg/val_list.txt
+  num_classes: 2
+  mode: val
+  transforms:
+    - type: Normalize
+
+optimizer:
+  type: sgd
+  momentum: 0.9
+  weight_decay: 4.0e-5
+
+lr_scheduler:
+  type: PolynomialDecay
+  learning_rate: 0.01
+  end_lr: 0
+  power: 0.9
+
+loss:
+  types:
+    - type: CrossEntropyLoss
+  coef: [1, 1, 1]
+
+
+model:
+  type: PPLiteSeg
+  backbone:
+    type: STDC2
+    pretrained: https://bj.bcebos.com/paddleseg/dygraph/PP_STDCNet2.tar.gz
--- a/configs/rtformer/README.md
+++ b/configs/rtformer/README.md
+# RTFormer: Efficient Design for Real-Time Semantic Segmentation with Transformer
+
+## Reference
+
+> Wang, Jian, Chenhui Gou, Qiman Wu, Haocheng Feng, Junyu Han, Errui Ding, and Jingdong Wang. "RTFormer: Efficient Design for Real-Time Semantic Segmentation with Transformer." arXiv preprint arXiv:2210.07124 (2022).
+
+## Performance
+
+### Cityscapes
+
+| Model | Backbone | Resolution | Training Iters | mIoU | mIoU (flip) | mIoU (ms+flip) | Links |
+|-|-|-|-|-|-|-|-|
+|RTFormer-Base|-|1024x512|120000|79.24%|79.80%|80.19%|[model](https://paddleseg.bj.bcebos.com/dygraph/cityscapes/rtformer_base_cityscapes_1024x512_120k/model.pdparams) \| [log]() \| [vdl]()|
+|RTFormer-Slim|-|1024x512|120000|76.31%|77.05%|77.58%|[model](https://paddleseg.bj.bcebos.com/dygraph/cityscapes/rtformer_slim_cityscapes_1024x512_120k/model.pdparams) \| [log]() \| [vdl]()|
+
+
+### ADE20k
+
+| Model | Backbone | Resolution | Training Iters | mIoU | mIoU (flip) | mIoU (ms+flip) | Links |
+|-|-|-|-|-|-|-|-|
+|RTFormer-Base|-|512x512|160000|42.02%|42.43%|42.72%|[model](https://paddleseg.bj.bcebos.com/dygraph/ade20k/rtformer_base_ade20k_512x512_160k/model.pdparams) \| [log]() \| [vdl]()|
+|RTFormer-Slim|-|512x512|160000|36.67%|37.32%|37.20%|[model](https://paddleseg.bj.bcebos.com/dygraph/ade20k/rtformer_slim_ade20k_512x512_160k/model.pdparams) \| [log]() \| [vdl]()|
--- a/configs/rtformer/rtformer_base_ade20k_512x512_160k.yml
+++ b/configs/rtformer/rtformer_base_ade20k_512x512_160k.yml
+_base_: '../_base_/ade20k.yml'
+
+batch_size: 4 # total batch size:  4 * 4
+iters: 160000
+
+train_dataset:
+  transforms:
+    - type: ResizeStepScaling
+      min_scale_factor: 0.5
+      max_scale_factor: 2.0
+      scale_step_size: 0.25
+    - type: RandomPaddingCrop
+      crop_size: [512, 512]
+    - type: RandomHorizontalFlip
+    - type: RandomDistort
+      brightness_range: 0.4
+      contrast_range: 0.4
+      saturation_range: 0.4
+    - type: Normalize
+      mean: [0.485, 0.456, 0.406]
+      std: [0.229, 0.224, 0.225]
+
+val_dataset:
+  transforms:
+    - type: Resize
+      target_size: [2048, 512]
+      keep_ratio: True
+      size_divisor: 32
+    - type: Normalize
+      mean: [0.485, 0.456, 0.406]
+      std: [0.229, 0.224, 0.225]
+  
+export:
+  transforms:
+    - type: Resize
+      target_size: [2048, 512]
+      keep_ratio: True
+      size_divisor: 32
+    - type: Normalize
+      mean: [0.485, 0.456, 0.406]
+      std: [0.229, 0.224, 0.225]
+
+optimizer:
+  _inherited_: False
+  type: AdamW
+  beta1: 0.9
+  beta2: 0.999
+  weight_decay: 0.05
+
+lr_scheduler:
+  _inherited_: False
+  type: PolynomialDecay
+  learning_rate: 1.0e-4
+  power: 1.
+  end_lr: 1.0e-7
+  warmup_iters: 1500
+  warmup_start_lr: 1.0e-6
+
+loss:
+  types:
+    - type: CrossEntropyLoss
+  coef: [1, 0.4]
+
+model:
+  type: RTFormer
+  base_channels: 64
+  head_channels: 128
+  drop_path_rate: 0.1
+  use_injection: [True, False]
+  pretrained: https://paddleseg.bj.bcebos.com/dygraph/backbone/rtformer_base_backbone_imagenet_pretrained.zip
--- a/configs/rtformer/rtformer_base_cityscapes_1024x512_120k.yml
+++ b/configs/rtformer/rtformer_base_cityscapes_1024x512_120k.yml
+_base_: '../_base_/cityscapes.yml'
+
+batch_size: 3 # total batch size:  4 * 3
+iters: 120000
+
+train_dataset:
+  transforms:
+    - type: ResizeStepScaling
+      min_scale_factor: 0.5
+      max_scale_factor: 2.0
+      scale_step_size: 0.25
+    - type: RandomPaddingCrop
+      crop_size: [1024, 512]
+    - type: RandomHorizontalFlip
+    - type: RandomDistort
+      brightness_range: 0.4
+      contrast_range: 0.4
+      saturation_range: 0.4
+    - type: Normalize
+      mean: [0.485, 0.456, 0.406]
+      std: [0.229, 0.224, 0.225]
+
+val_dataset:
+  transforms:
+    - type: Resize
+      target_size: [2048, 1024]
+      keep_ratio: True
+    - type: Normalize
+      mean: [0.485, 0.456, 0.406]
+      std: [0.229, 0.224, 0.225]
+  
+export:
+  transforms:
+    - type: Resize
+      target_size: [2048, 512]
+      keep_ratio: True
+    - type: Normalize
+      mean: [0.485, 0.456, 0.406]
+      std: [0.229, 0.224, 0.225]
+
+optimizer:
+  _inherited_: False
+  type: AdamW
+  beta1: 0.9
+  beta2: 0.999
+  weight_decay: 0.0125
+
+lr_scheduler:
+  _inherited_: False
+  type: PolynomialDecay
+  learning_rate: 4.0e-4
+  power: 1.
+  end_lr: 1.0e-6
+  warmup_iters: 1500
+  warmup_start_lr: 1.0e-6
+
+loss:
+  types:
+    - type: CrossEntropyLoss
+  coef: [1, 0.4]
+
+model:
+  type: RTFormer
+  base_channels: 64
+  head_channels: 128
+  use_injection: [True, False]
+  pretrained: https://paddleseg.bj.bcebos.com/dygraph/backbone/rtformer_base_backbone_imagenet_pretrained.zip
--- a/configs/rtformer/rtformer_slim_ade20k_512x512_160k.yml
+++ b/configs/rtformer/rtformer_slim_ade20k_512x512_160k.yml
+_base_: '../_base_/ade20k.yml'
+
+batch_size: 4 # total batch size:  4 * 4
+iters: 160000
+
+train_dataset:
+  transforms:
+    - type: ResizeStepScaling
+      min_scale_factor: 0.5
+      max_scale_factor: 2.0
+      scale_step_size: 0.25
+    - type: RandomPaddingCrop
+      crop_size: [512, 512]
+    - type: RandomHorizontalFlip
+    - type: RandomDistort
+      brightness_range: 0.4
+      contrast_range: 0.4
+      saturation_range: 0.4
+    - type: Normalize
+      mean: [0.485, 0.456, 0.406]
+      std: [0.229, 0.224, 0.225]
+
+val_dataset:
+  transforms:
+    - type: Resize
+      target_size: [2048, 512]
+      keep_ratio: True
+      size_divisor: 32
+    - type: Normalize
+      mean: [0.485, 0.456, 0.406]
+      std: [0.229, 0.224, 0.225]
+  
+export:
+  transforms:
+    - type: Resize
+      target_size: [2048, 512]
+      keep_ratio: True
+      size_divisor: 32
+    - type: Normalize
+      mean: [0.485, 0.456, 0.406]
+      std: [0.229, 0.224, 0.225]
+
+optimizer:
+  _inherited_: False
+  type: AdamW
+  beta1: 0.9
+  beta2: 0.999
+  weight_decay: 0.05
+
+lr_scheduler:
+  _inherited_: False
+  type: PolynomialDecay
+  learning_rate: 1.0e-4
+  power: 1.
+  end_lr: 1.0e-7
+  warmup_iters: 1500
+  warmup_start_lr: 1.0e-6
+
+loss:
+  types:
+    - type: CrossEntropyLoss
+  coef: [1, 0.4]
+
+model:
+  type: RTFormer
+  base_channels: 32
+  head_channels: 64
+  drop_path_rate: 0.1
+  cross_size: 8
+  use_injection: [True, False]
+  pretrained: https://paddleseg.bj.bcebos.com/dygraph/backbone/rtformer_slim_backbone_imagenet_pretrained.zip
--- a/configs/rtformer/rtformer_slim_cityscapes_1024x512_120k.yml
+++ b/configs/rtformer/rtformer_slim_cityscapes_1024x512_120k.yml
+_base_: '../_base_/cityscapes.yml'
+
+batch_size: 3 # total batch size:  4 * 3
+iters: 120000
+
+train_dataset:
+  transforms:
+    - type: ResizeStepScaling
+      min_scale_factor: 0.5
+      max_scale_factor: 2.0
+      scale_step_size: 0.25
+    - type: RandomPaddingCrop
+      crop_size: [1024, 512]
+    - type: RandomHorizontalFlip
+    - type: RandomDistort
+      brightness_range: 0.4
+      contrast_range: 0.4
+      saturation_range: 0.4
+    - type: Normalize
+      mean: [0.485, 0.456, 0.406]
+      std: [0.229, 0.224, 0.225]
+
+val_dataset:
+  transforms:
+    - type: Resize
+      target_size: [2048, 1024]
+      keep_ratio: True
+    - type: Normalize
+      mean: [0.485, 0.456, 0.406]
+      std: [0.229, 0.224, 0.225]
+  
+export:
+  transforms:
+    - type: Resize
+      target_size: [2048, 512]
+      keep_ratio: True
+    - type: Normalize
+      mean: [0.485, 0.456, 0.406]
+      std: [0.229, 0.224, 0.225]
+
+optimizer:
+  _inherited_: False
+  type: AdamW
+  beta1: 0.9
+  beta2: 0.999
+  weight_decay: 0.0125
+
+lr_scheduler:
+  _inherited_: False
+  type: PolynomialDecay
+  learning_rate: 4.0e-4
+  power: 1.
+  end_lr: 1.0e-6
+  warmup_iters: 1500
+  warmup_start_lr: 1.0e-6
+
+loss:
+  types:
+    - type: CrossEntropyLoss
+  coef: [1, 0.4]
+
+model:
+  type: RTFormer
+  base_channels: 32
+  head_channels: 64
+  use_injection: [True, True]
+  pretrained: https://paddleseg.bj.bcebos.com/dygraph/backbone/rtformer_slim_backbone_imagenet_pretrained.zip
\ No newline at end of file
--- a/configs/segformer/README.md
+++ b/configs/segformer/README.md
+# SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers
+
+## Reference
+
+> Xie, Enze, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, and Ping Luo. "SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers." arXiv preprint arXiv:2105.15203 (2021).
+
+## Performance
+
+### Cityscapes
+
+| Model | Backbone | Resolution | Training Iters | mIoU(slice) | mIoU (flip) | mIoU (ms+flip) | Links |
+|-|-|-|-|-|-|-|-|
+|SegFormer_B0|-|1024x1024|160000|76.73%|77.16%|-|[model](https://bj.bcebos.com/paddleseg/dygraph/cityscapes/segformer_b0_cityscapes_1024x1024_160k/model.pdparams) \| [log](https://bj.bcebos.com/paddleseg/dygraph/cityscapes/segformer_b0_cityscapes_1024x1024_160k/train.log) \| [vdl](https://paddlepaddle.org.cn/paddle/visualdl/service/app?id=227e067add44d44383c402ec5aead11b)|
+|SegFormer_B1|-|1024x1024|160000|78.35%|78.64%|-|[model](https://bj.bcebos.com/paddleseg/dygraph/cityscapes/segformer_b1_cityscapes_1024x1024_160k/model.pdparams) \| [log](https://bj.bcebos.com/paddleseg/dygraph/cityscapes/segformer_b1_cityscapes_1024x1024_160k/train.log) \| [vdl](https://paddlepaddle.org.cn/paddle/visualdl/service/app?id=a0f4e8eacf346826e3150989b6a9f849)|
+|SegFormer_B2|-|1024x1024|160000|81.60%|81.82%|-|[model](https://bj.bcebos.com/paddleseg/dygraph/cityscapes/segformer_b2_cityscapes_1024x1024_160k/model.pdparams) \| [log](https://bj.bcebos.com/paddleseg/dygraph/cityscapes/segformer_b2_cityscapes_1024x1024_160k/train.log) \| [vdl](https://paddlepaddle.org.cn/paddle/visualdl/service/app?id=734c0d99d858d0db7ff58f03d18289fe)|
+|SegFormer_B3|-|1024x1024|160000|82.47%|82.60%|-|[model](https://bj.bcebos.com/paddleseg/dygraph/cityscapes/segformer_b3_cityscapes_1024x1024_160k/model.pdparams) \| [log](https://bj.bcebos.com/paddleseg/dygraph/cityscapes/segformer_b3_cityscapes_1024x1024_160k/train.log) \| [vdl](https://paddlepaddle.org.cn/paddle/visualdl/service/app?id=406282a64c45d008bf4445c5669d6579)|
+|SegFormer_B4|-|1024x1024|160000|82.38%|82.59%|-|[model](https://bj.bcebos.com/paddleseg/dygraph/cityscapes/segformer_b4_cityscapes_1024x1024_160k/model.pdparams) \| [log](https://bj.bcebos.com/paddleseg/dygraph/cityscapes/segformer_b4_cityscapes_1024x1024_160k/train.log) \| [vdl](https://paddlepaddle.org.cn/paddle/visualdl/service/app?id=dc51a262eb8be9273970354ed445e760)|
+|SegFormer_B5|-|1024x1024|160000|82.58%|82.82%|-|[model](https://bj.bcebos.com/paddleseg/dygraph/cityscapes/segformer_b5_cityscapes_1024x1024_160k/model.pdparams) \| [log](https://bj.bcebos.com/paddleseg/dygraph/cityscapes/segformer_b5_cityscapes_1024x1024_160k/train.log) \| [vdl](https://paddlepaddle.org.cn/paddle/visualdl/service/app?id=306d042a8e4d82ccceabd988a478a2f8)|
--- a/configs/segformer/segformer_b0_cityscapes_1024x1024_160k.yml
+++ b/configs/segformer/segformer_b0_cityscapes_1024x1024_160k.yml
+_base_: '../_base_/cityscapes_1024x1024.yml'
+
+batch_size: 2
+iters: 160000
+
+model:
+  type: SegFormer
+  backbone:
+    type: MixVisionTransformer_B0
+    pretrained: https://bj.bcebos.com/paddleseg/dygraph/backbone/mix_vision_transformer_b0.tar.gz
+  embedding_dim: 256
+  num_classes: 19
+
+optimizer:
+  _inherited_: False
+  type: AdamW
+  beta1: 0.9
+  beta2: 0.999
+  weight_decay: 0.01
+
+lr_scheduler:
+  type: PolynomialDecay
+  learning_rate: 0.00006
+  power: 1
+
+loss:
+  types:
+    - type: CrossEntropyLoss
+  coef: [1]
+
+test_config:
+    is_slide: True
+    crop_size: [1024, 1024]
+    stride: [768, 768]
--- a/configs/segformer/segformer_b0_cityscapes_1024x512_160k.yml
+++ b/configs/segformer/segformer_b0_cityscapes_1024x512_160k.yml
+_base_: '../_base_/cityscapes.yml'
+
+batch_size: 1
+iters: 160000
+
+model:
+  type: SegFormer
+  backbone:
+    type: MixVisionTransformer_B0
+    pretrained: https://bj.bcebos.com/paddleseg/dygraph/backbone/mix_vision_transformer_b0.tar.gz
+  embedding_dim: 256
+  num_classes: 19
+
+optimizer:
+  _inherited_: False
+  type: AdamW
+  beta1: 0.9
+  beta2: 0.999
+  weight_decay: 0.01
+
+lr_scheduler:
+  type: PolynomialDecay
+  learning_rate: 0.00006
+  power: 1
+
+loss:
+  types:
+    - type: CrossEntropyLoss
+  coef: [1]
--- a/configs/segformer/segformer_b1_cityscapes_1024x1024_160k.yml
+++ b/configs/segformer/segformer_b1_cityscapes_1024x1024_160k.yml
+_base_: '../_base_/cityscapes_1024x1024.yml'
+
+batch_size: 2
+iters: 160000
+
+model:
+  type: SegFormer
+  backbone:
+    type: MixVisionTransformer_B1
+    pretrained: https://bj.bcebos.com/paddleseg/dygraph/backbone/mix_vision_transformer_b1.tar.gz
+  embedding_dim: 256
+  num_classes: 19
+
+optimizer:
+  _inherited_: False
+  type: AdamW
+  beta1: 0.9
+  beta2: 0.999
+  weight_decay: 0.01
+
+lr_scheduler:
+  type: PolynomialDecay
+  learning_rate: 0.00006
+  power: 1
+
+loss:
+  types:
+    - type: CrossEntropyLoss
+  coef: [1]
+
+test_config:
+    is_slide: True
+    crop_size: [1024, 1024]
+    stride: [768, 768]
--- a/configs/segformer/segformer_b1_cityscapes_1024x512_160k.yml
+++ b/configs/segformer/segformer_b1_cityscapes_1024x512_160k.yml
+_base_: '../_base_/cityscapes.yml'
+
+batch_size: 1
+iters: 160000
+
+model:
+  type: SegFormer
+  backbone:
+    type: MixVisionTransformer_B1
+    pretrained: https://bj.bcebos.com/paddleseg/dygraph/backbone/mix_vision_transformer_b1.tar.gz
+  embedding_dim: 256
+  num_classes: 19
+
+optimizer:
+  _inherited_: False
+  type: AdamW
+  beta1: 0.9
+  beta2: 0.999
+  weight_decay: 0.01
+
+lr_scheduler:
+  type: PolynomialDecay
+  learning_rate: 0.00006
+  power: 1
+
+loss:
+  types:
+    - type: CrossEntropyLoss
+  coef: [1]
--- a/configs/segformer/segformer_b2_cityscapes_1024x1024_160k.yml
+++ b/configs/segformer/segformer_b2_cityscapes_1024x1024_160k.yml
+_base_: '../_base_/cityscapes_1024x1024.yml'
+
+batch_size: 2
+iters: 160000
+
+model:
+  type: SegFormer
+  backbone:
+    type: MixVisionTransformer_B2
+    pretrained: https://bj.bcebos.com/paddleseg/dygraph/backbone/mix_vision_transformer_b2.tar.gz
+  embedding_dim: 768
+  num_classes: 19
+
+optimizer:
+  _inherited_: False
+  type: AdamW
+  beta1: 0.9
+  beta2: 0.999
+  weight_decay: 0.01
+
+lr_scheduler:
+  type: PolynomialDecay
+  learning_rate: 0.00006
+  power: 1
+
+loss:
+  types:
+    - type: CrossEntropyLoss
+  coef: [1]
+
+test_config:
+    is_slide: True
+    crop_size: [1024, 1024]
+    stride: [768, 768]
--- a/configs/segformer/segformer_b2_cityscapes_1024x512_160k.yml
+++ b/configs/segformer/segformer_b2_cityscapes_1024x512_160k.yml
+_base_: '../_base_/cityscapes.yml'
+
+batch_size: 1
+iters: 160000
+
+model:
+  type: SegFormer
+  backbone:
+    type: MixVisionTransformer_B2
+    pretrained: https://bj.bcebos.com/paddleseg/dygraph/backbone/mix_vision_transformer_b2.tar.gz
+  embedding_dim: 768
+  num_classes: 19
+
+optimizer:
+  _inherited_: False
+  type: AdamW
+  beta1: 0.9
+  beta2: 0.999
+  weight_decay: 0.01
+
+lr_scheduler:
+  type: PolynomialDecay
+  learning_rate: 0.00006
+  power: 1
+
+loss:
+  types:
+    - type: CrossEntropyLoss
+  coef: [1]
--- a/configs/segformer/segformer_b3_cityscapes_1024x1024_160k.yml
+++ b/configs/segformer/segformer_b3_cityscapes_1024x1024_160k.yml
+_base_: '../_base_/cityscapes_1024x1024.yml'
+
+batch_size: 2
+iters: 160000
+
+model:
+  type: SegFormer
+  backbone:
+    type: MixVisionTransformer_B3
+    pretrained: https://bj.bcebos.com/paddleseg/dygraph/backbone/mix_vision_transformer_b3.tar.gz
+  embedding_dim: 768
+  num_classes: 19
+
+optimizer:
+  _inherited_: False
+  type: AdamW
+  beta1: 0.9
+  beta2: 0.999
+  weight_decay: 0.01
+
+lr_scheduler:
+  type: PolynomialDecay
+  learning_rate: 0.00006
+  power: 1
+
+loss:
+  types:
+    - type: CrossEntropyLoss
+  coef: [1]
+
+test_config:
+    is_slide: True
+    crop_size: [1024, 1024]
+    stride: [768, 768]
--- a/configs/segformer/segformer_b3_cityscapes_1024x512_160k.yml
+++ b/configs/segformer/segformer_b3_cityscapes_1024x512_160k.yml
+_base_: '../_base_/cityscapes.yml'
+
+batch_size: 1
+iters: 160000
+
+model:
+  type: SegFormer
+  backbone:
+    type: MixVisionTransformer_B3
+    pretrained: https://bj.bcebos.com/paddleseg/dygraph/backbone/mix_vision_transformer_b3.tar.gz
+  embedding_dim: 768
+  num_classes: 19
+
+optimizer:
+  _inherited_: False
+  type: AdamW
+  beta1: 0.9
+  beta2: 0.999
+  weight_decay: 0.01
+
+lr_scheduler:
+  type: PolynomialDecay
+  learning_rate: 0.00006
+  power: 1
+
+loss:
+  types:
+    - type: CrossEntropyLoss
+  coef: [1]
--- a/configs/segformer/segformer_b4_cityscapes_1024x1024_160k.yml
+++ b/configs/segformer/segformer_b4_cityscapes_1024x1024_160k.yml
+_base_: '../_base_/cityscapes_1024x1024.yml'
+
+batch_size: 2
+iters: 160000
+
+model:
+  type: SegFormer
+  backbone:
+    type: MixVisionTransformer_B4
+    pretrained: https://bj.bcebos.com/paddleseg/dygraph/backbone/mix_vision_transformer_b4.tar.gz
+  embedding_dim: 768
+  num_classes: 19
+
+optimizer:
+  _inherited_: False
+  type: AdamW
+  beta1: 0.9
+  beta2: 0.999
+  weight_decay: 0.01
+
+lr_scheduler:
+  type: PolynomialDecay
+  learning_rate: 0.00006
+  power: 1
+
+loss:
+  types:
+    - type: CrossEntropyLoss
+  coef: [1]
+
+test_config:
+    is_slide: True
+    crop_size: [1024, 1024]
+    stride: [768, 768]
--- a/configs/segformer/segformer_b4_cityscapes_1024x512_160k.yml
+++ b/configs/segformer/segformer_b4_cityscapes_1024x512_160k.yml
+_base_: '../_base_/cityscapes.yml'
+
+batch_size: 1
+iters: 160000
+
+model:
+  type: SegFormer
+  backbone:
+    type: MixVisionTransformer_B4
+    pretrained: https://bj.bcebos.com/paddleseg/dygraph/backbone/mix_vision_transformer_b4.tar.gz
+  embedding_dim: 768
+  num_classes: 19
+
+optimizer:
+  _inherited_: False
+  type: AdamW
+  beta1: 0.9
+  beta2: 0.999
+  weight_decay: 0.01
+
+lr_scheduler:
+  type: PolynomialDecay
+  learning_rate: 0.00006
+  power: 1
+
+loss:
+  types:
+    - type: CrossEntropyLoss
+  coef: [1]