add new model

configs/smrt/ocrnet_hrnetw18.yml

+_base_: './base_cfg.yml'
+
+model:
+  type: OCRNet
+  backbone:
+    type: HRNet_W18
+    pretrained: https://bj.bcebos.com/paddleseg/dygraph/hrnet_w18_ssld.tar.gz
+  backbone_indices: [0]
+
+loss:
+  types:
+    - type: MixedLoss
+      losses:
+        - type: OhemCrossEntropyLoss
+          min_kept: 65000
+        - type: LovaszSoftmaxLoss
+      coef: [0.8, 0.2]
+  coef: [1, 0.4]
\ No newline at end of file

configs/smrt/pp_liteseg_stdc1.yml

+_base_: './base_cfg.yml'
+
+model:
+  type: PPLiteSeg
+  backbone:
+    type: STDC1
+    pretrained: https://bj.bcebos.com/paddleseg/dygraph/PP_STDCNet1.tar.gz
+  arm_out_chs: [32, 64, 128]
+  seg_head_inter_chs: [32, 64, 64]
+
+loss:
+  types:
+    - type: MixedLoss
+      losses:
+        - type: OhemCrossEntropyLoss
+          min_kept: 65000
+        - type: LovaszSoftmaxLoss
+      coef: [0.8, 0.2]
+  coef: [1, 1, 1]
\ No newline at end of file

configs/smrt/pp_liteseg_stdc2.yml

+_base_: './base_cfg.yml'
+
+model:
+  type: PPLiteSeg
+  backbone:
+    type: STDC2
+    pretrained: https://bj.bcebos.com/paddleseg/dygraph/PP_STDCNet2.tar.gz
+
+loss:
+  types:
+    - type: MixedLoss
+      losses:
+        - type: OhemCrossEntropyLoss
+          min_kept: 65000
+        - type: LovaszSoftmaxLoss
+      coef: [0.8, 0.2]
+  coef: [1, 1, 1]
\ No newline at end of file

configs/smrt/sfnet_resnet18_os8.yml

+_base_: './base_cfg.yml'
+
+model:
+  type: SFNet
+  backbone:
+    type: ResNet18_vd
+    output_stride: 8
+    pretrained: https://bj.bcebos.com/paddleseg/dygraph/resnet18_vd_ssld_v2.tar.gz
+  backbone_indices: [0, 1, 2, 3]
+
+loss:
+  types:
+    - type: MixedLoss
+      losses:
+        - type: OhemCrossEntropyLoss
+          min_kept: 65000
+        - type: LovaszSoftmaxLoss
+      coef: [0.8, 0.2]
+  coef: [1]
\ No newline at end of file

configs/stdcseg/README.md

+# Rethinking BiSeNet For Real-time Semantic Segmentation
+
+## Reference
+
+> Fan, Mingyuan, Shenqi Lai, Junshi Huang, Xiaoming Wei, Zhenhua Chai, Junfeng Luo, and Xiaolin Wei. "Rethinking BiSeNet For Real-time Semantic Segmentation." In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 9716-9725. 2021.
+
+
+## Performance
+
+### CityScapes
+
+| Model | Backbone | Resolution | Training Iters | mIoU | mIoU (flip) | mIoU (ms+flip) | Links |
+|---|---|---|---|---|---|---|---|
+|STDC1-Seg50|STDC1|1024x512|80000|74.74%|75.71%|76.64%|[model](https://bj.bcebos.com/paddleseg/dygraph/cityscapes/stdc1_seg_cityscapes_1024x512_80k/model.pdparams) \| [log](https://bj.bcebos.com/paddleseg/dygraph/cityscapes/stdc1_seg_cityscapes_1024x512_80k/train.log) \| [vdl](https://paddlepaddle.org.cn/paddle/visualdl/service/app?id=f450fdef2b3b02574e3eb293242d1fbd) |
+|STDC2-Seg50|STDC2|1024x512|80000|77.60%|78.32%|78.83%|[model](https://bj.bcebos.com/paddleseg/dygraph/cityscapes/stdc2_seg_cityscapes_1024x512_80k/model.pdparams) \| [log](https://bj.bcebos.com/paddleseg/dygraph/cityscapes/stdc2_seg_cityscapes_1024x512_80k/train.log) \| [vdl](https://paddlepaddle.org.cn/paddle/visualdl/service/app?id=1f9c39f10e94327803faae96a516a7a6) |
+
+### Pascal VOC 2012 + Aug
+
+| Model | Backbone | Resolution | Training Iters | mIoU | mIoU (flip) | mIoU (ms+flip) | Links |
+|:-:|:-:|:-:|:-:|:-:|:-:|:-:|:-:|
+|STDC1-Seg50|STDC1|512x512|40000|68.06%|68.48%|69.04%|[model](https://bj.bcebos.com/paddleseg/dygraph/pascal_voc12/stdc1_seg_voc12aug_512x512_40k/model.pdparams) \| [log](https://bj.bcebos.com/paddleseg/dygraph/pascal_voc12/stdc1_seg_voc12aug_512x512_40k/train.log) \| [vdl](https://paddlepaddle.org.cn/paddle/visualdl/service/app?id=86061db4140c995922b033f96945d3da) |
+|STDC2-Seg50|STDC2|512x512|40000|68.98%|70.07%|69.99%|[model](https://bj.bcebos.com/paddleseg/dygraph/pascal_voc12/stdc2_seg_voc12aug_512x512_40k/model.pdparams) \| [log](https://bj.bcebos.com/paddleseg/dygraph/pascal_voc12/stdc2_seg_voc12aug_512x512_40k/train.log) \| [vdl](https://paddlepaddle.org.cn/paddle/visualdl/service/app?id=46d5d3cead36ee9d16df1d06b121b3bc) |

configs/stdcseg/stdc1_seg_cityscapes_1024x512_80k.yml

+_base_: '../_base_/cityscapes.yml'
+
+batch_size: 12
+iters: 80000
+
+model:
+  type: STDCSeg
+  backbone:
+    type: STDC1
+    pretrained: https://bj.bcebos.com/paddleseg/dygraph/STDCNet1.tar.gz
+  pretrained: null
+
+loss:
+  types:
+    - type: OhemCrossEntropyLoss
+    - type: OhemCrossEntropyLoss
+    - type: OhemCrossEntropyLoss
+    - type: DetailAggregateLoss
+  coef: [1, 1, 1, 1]

configs/stdcseg/stdc1_seg_voc12aug_512x512_40k.yml

+_base_: '../_base_/pascal_voc12aug.yml'
+
+
+model:
+  type: STDCSeg
+  backbone:
+    type: STDC1
+    pretrained: https://bj.bcebos.com/paddleseg/dygraph/STDCNet1.tar.gz
+  pretrained: null
+
+loss:
+  types:
+    - type: OhemCrossEntropyLoss
+    - type: OhemCrossEntropyLoss
+    - type: OhemCrossEntropyLoss
+    - type: DetailAggregateLoss
+  coef: [1, 1, 1, 1]

configs/stdcseg/stdc2_seg_cityscapes_1024x512_80k.yml

+_base_: 'stdc1_seg_cityscapes_1024x512_80k.yml'
+
+
+model:
+  backbone:
+    type: STDC2
+    pretrained: https://bj.bcebos.com/paddleseg/dygraph/STDCNet2.tar.gz

configs/stdcseg/stdc2_seg_voc12aug_512x512_40k.yml

+_base_: 'stdc1_seg_voc12aug_512x512_40k.yml'
+
+
+model:
+  backbone:
+    type: STDC2
+    pretrained: https://bj.bcebos.com/paddleseg/dygraph/STDCNet2.tar.gz

configs/topformer/README.md

+# TopFormer: Token Pyramid Transformer for Mobile Semantic Segmentation
+
+## Reference
+
+> Zhang, Wenqiang, Zilong Huang, Guozhong Luo, Tao Chen, Xinggang Wang, Wenyu Liu, Gang Yu,and Chunhua Shen. "TopFormer: Token Pyramid Transformer for Mobile Semantic Segmentation." In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 12083-12093. 2022.
+
+
+## Performance
+
+### ADE20k
+
+| Model | Backbone | Resolution | Training Iters | mIoU | mIoU (flip) | mIoU (ms+flip) | Links |
+|---|---|---|---|---|---|---|---|
+|TopFormer-Base |topformer|512x512|160000| 38.28% | 38.59% | - |[model](https://paddleseg.bj.bcebos.com/dygraph/ade20k/topformer_base_ade20k_512x512_160k/model.pdparams) \| [log](https://paddleseg.bj.bcebos.com/dygraph/ade20k/topformer_base_ade20k_512x512_160k/log_train.txt) \| [vdl](https://paddlepaddle.org.cn/paddle/visualdl/service/app?id=1530ee894b33a363677472fdcae5d13a) |
+|TopFormer-Small|topformer|512x512|160000| 35.60% | 35.83% | - |[model](https://paddleseg.bj.bcebos.com/dygraph/ade20k/topformer_small_ade20k_512x512_160k/model.pdparams) \| [log](https://paddleseg.bj.bcebos.com/dygraph/ade20k/topformer_small_ade20k_512x512_160k/log_train.txt) \| [vdl](https://paddlepaddle.org.cn/paddle/visualdl/service/app?id=c6070db4366510a20d47fb4645797a27) |
+|TopFormer-Tiny |topformer|512x512|160000| 32.49% | 32.75% | - |[model](https://paddleseg.bj.bcebos.com/dygraph/ade20k/topformer_tiny_ade20k_512x512_160k/model.pdparams) \| [log](https://paddleseg.bj.bcebos.com/dygraph/ade20k/topformer_tiny_ade20k_512x512_160k/log_train.txt) \| [vdl](https://paddlepaddle.org.cn/paddle/visualdl/service/app?id=462723f835db022d3eba1b4db87350e3) |
+
+
+Note that, the input resulution of TopFormer should be a multiple of 32.

configs/topformer/topformer_base_ade20k_512x512_160k.yml

+_base_: '../_base_/ade20k.yml'
+
+batch_size: 4  # total batch size is 16
+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
+  weight_decay: 0.01
+
+lr_scheduler:
+  type: PolynomialDecay
+  learning_rate: 0.0012
+  end_lr: 0
+  power: 1.0
+  warmup_iters: 1500
+  warmup_start_lr: 1.0e-6
+
+loss:
+  types:
+    - type: CrossEntropyLoss
+  coef: [1]
+
+model:
+  type: TopFormer
+  backbone:
+    type: TopTransformer_Base
+    lr_mult: 0.1
+    pretrained: https://paddleseg.bj.bcebos.com/dygraph/backbone/topformer_base_imagenet_pretrained.zip
\ No newline at end of file

configs/topformer/topformer_small_ade20k_512x512_160k.yml

+_base_: '../_base_/ade20k.yml'
+
+batch_size: 4  # total batch size is 16
+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
+  weight_decay: 0.01
+
+lr_scheduler:
+  type: PolynomialDecay
+  learning_rate: 0.0012
+  end_lr: 0
+  power: 1.0
+  warmup_iters: 1500
+  warmup_start_lr: 1.0e-6
+
+loss:
+  types:
+    - type: CrossEntropyLoss
+  coef: [1]
+
+model:
+  type: TopFormer
+  backbone:
+    type: TopTransformer_Small
+    lr_mult: 0.1
+    pretrained: https://paddleseg.bj.bcebos.com/dygraph/backbone/topformer_small_imagenet_pretrained.zip
\ No newline at end of file

configs/topformer/topformer_tiny_ade20k_512x512_160k.yml

+_base_: '../_base_/ade20k.yml'
+
+batch_size: 4  # total batch size is 16
+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
+  weight_decay: 0.01
+
+lr_scheduler:
+  type: PolynomialDecay
+  learning_rate: 0.0012
+  end_lr: 0
+  power: 1.0
+  warmup_iters: 1500
+  warmup_start_lr: 1.0e-6
+
+loss:
+  types:
+    - type: CrossEntropyLoss
+  coef: [1]
+
+model:
+  type: TopFormer
+  head_use_dw: True
+  backbone:
+    type: TopTransformer_Tiny
+    lr_mult: 0.1
+    pretrained: https://paddleseg.bj.bcebos.com/dygraph/backbone/topformer_tiny_imagenet_pretrained.zip
\ No newline at end of file

configs/u2net/README.md

+# U2-Net: Going deeper with nested U-structure for salient object detection
+
+## Reference
+> Qin, Xuebin, Zichen Zhang, Chenyang Huang, Masood Dehghan, Osmar R. Zaiane, and Martin Jagersand. "U2-Net: Going deeper with nested U-structure for salient object detection." Pattern Recognition 106 (2020): 107404.

configs/u2net/u2net_cityscapes_1024x512_160k.yml

+_base_: '../_base_/cityscapes.yml'
+
+batch_size: 4
+iters: 160000
+
+model:
+  type: U2Net
+  num_classes: 19
+  pretrained: Null
+
+loss:
+  coef: [1, 1, 1, 1, 1, 1, 1]

configs/u2net/u2netp_cityscapes_1024x512_160k.yml

+_base_: '../_base_/cityscapes.yml'
+
+batch_size: 4
+iters: 160000
+
+model:
+  type: U2Netp
+  num_classes: 19
+  pretrained: Null
+
+loss:
+  coef: [1, 1, 1, 1, 1, 1, 1]

configs/uhrnet/README.md

+# U-HRNet: Delving into Improving Semantic Representation of High Resolution Network for Dense Prediction
+
+## Reference
+> Jian Wang, Xiang Long, Guowei Chen, Zewu Wu, Zeyu Chen, Errui Ding et al. "U-HRNet: Delving into Improving Semantic Representation of High Resolution Network for Dense Prediction" arXiv preprint arXiv:2210.07140 (2022).
+
+## Performance
+
+### Cityscapes
+
+| Model | Backbone | Resolution | Training Iters | mIoU | mIoU (flip) | mIoU (ms+flip) | Links |
+|:-:|:-:|:-:|:-:|:-:|:-:|:-:|:-:|
+|FCN|UHRNet_W18_small|1024x512|80000|77.66%|78.26%|78.47%|[model](https://bj.bcebos.com/paddleseg/dygraph/cityscapes/fcn_uhrnetw18_small_cityscapes_1024x512_80k/model.pdparams) \| [log](https://bj.bcebos.com/paddleseg/dygraph/cityscapes/fcn_uhrnetw18_small_cityscapes_1024x512_80k/train.log) \| [vdl](https://paddlepaddle.org.cn/paddle/visualdl/service/app?id=9cb0e961bc1f89d3484190f9d4de550b)|
+|FCN|UHRNet_W18_small|1024x512|120000|78.39%|79.09%|79.03%|[model](https://bj.bcebos.com/paddleseg/dygraph/cityscapes/fcn_uhrnetw18_small_cityscapes_1024x512_120k_bs3/model.pdparams) \| [log](https://bj.bcebos.com/paddleseg/dygraph/cityscapes/fcn_uhrnetw18_small_cityscapes_1024x512_120k_bs3/train.log) \| [vdl](https://paddlepaddle.org.cn/paddle/visualdl/service/app?id=6f6c41e46cf8b26d3a941bf7e09698f8)|
+|FCN|UHRNet_W48|1024x512|80000|81.28%|81.76%|81.48%|[model](https://bj.bcebos.com/paddleseg/dygraph/cityscapes/fcn_uhrnetw48_cityscapes_1024x512_80k/model.pdparams) \| [log](https://bj.bcebos.com/paddleseg/dygraph/cityscapes/fcn_uhrnetw48_cityscapes_1024x512_80k/train.log) \| [vdl](https://paddlepaddle.org.cn/paddle/visualdl/service/app?id=1c2fbc3a5558d530c2a1fc8c2cd34da5)|
+|FCN|UHRNet_W48|1024x512|120000|81.91%|82.39%|82.28%|[model](https://bj.bcebos.com/paddleseg/dygraph/cityscapes/fcn_uhrnetw48_cityscapes_1024x512_120k_bs3/model.pdparams) \| [log](https://bj.bcebos.com/paddleseg/dygraph/cityscapes/fcn_uhrnetw48_cityscapes_1024x512_120k_bs3/train.log) \| [vdl](https://paddlepaddle.org.cn/paddle/visualdl/service/app?id=a94e548519f9487c435530532f7a027c)|

configs/uhrnet/fcn_uhrnetw18_small_cityscapes_1024x512_120k_bs3.yml

+_base_: './fcn_uhrnetw18_small_cityscapes_1024x512_80k.yml'
+
+batch_size: 3
+iters: 120000
+

configs/uhrnet/fcn_uhrnetw18_small_cityscapes_1024x512_80k.yml

+_base_: '../_base_/cityscapes.yml'
+
+model:
+  type: FCN
+  backbone:
+    type: UHRNet_W18_Small
+    align_corners: False
+    pretrained: https://bj.bcebos.com/paddleseg/dygraph/backbone/uhrnetw18_small_imagenet.tar.gz
+  num_classes: 19
+  pretrained: Null
+  backbone_indices: [-1]
+
+optimizer:
+  weight_decay: 0.0005

configs/uhrnet/fcn_uhrnetw48_cityscapes_1024x512_120k_bs3.yml

+_base_: './fcn_uhrnetw48_cityscapes_1024x512_80k.yml'
+
+batch_size: 3
+iters: 120000
\ No newline at end of file