batch_size: 1
iters: 80000

model:
  type: SegFormer
  backbone:
    type: MixVisionTransformer_B1
    pretrained: https://bj.bcebos.com/paddleseg/dygraph/backbone/mix_vision_transformer_b1.tar.gz
  embedding_dim: 256

train_dataset:
  type: Dataset
  dataset_root: datasets/Cityscapes
  train_path: datasets/Cityscapes/train.txt
  num_classes: 19
  transforms:
    - type: ResizeStepScaling
      min_scale_factor: 0.5
      max_scale_factor: 2.0
      scale_step_size: 0.25
    - type: RandomPaddingCrop
      crop_size: [1024, 1024]
    - type: RandomHorizontalFlip
    - type: RandomDistort
      brightness_range: 0.4
      contrast_range: 0.4
      saturation_range: 0.4
    - type: Normalize
  mode: train

val_dataset:
  type: Dataset
  dataset_root: datasets/Cityscapes
  val_path: datasets/Cityscapes/val.txt
  num_classes: 19
  transforms:
    - type: Normalize
  mode: val

optimizer:
  type: AdamW
  beta1: 0.9
  beta2: 0.999
  weight_decay: 0.01

lr_scheduler:
  type: PolynomialDecay
  learning_rate: 0.00006
  power: 1
  end_lr: 0

loss:
  types:
    - type: CrossEntropyLoss
  coef: [1]