Skip to content

Latest commit

 

History

History
155 lines (140 loc) · 5.64 KB

configs.md

File metadata and controls

155 lines (140 loc) · 5.64 KB
Raw

Configuration items

data_cfg

  • Data configurations
  • Args
    • dataset_name: support CASIA-B, OUMVLP, GREW and Gait3D now.
    • dataset_root: The path of storing your dataset.
    • num_workers: The number of workers to collect data.
    • dataset_partition: The path of storing your dataset partition file. It splits the dataset to two parts, including train set and test set.
    • cache: If True, load all data to memory during building dataset.
    • test_dataset_name: The name of test dataset.
    • frame_threshold: The threshold of the sequence length, default: 0

loss_cfg

  • Loss function
  • Args
    • type: Loss function type, support TripletLoss ,SupConLoss,SupConLoss_LP .
    • loss_term_weight: loss weight.
    • log_prefix: the prefix of loss log.

optimizer_cfg

  • Optimizer
  • Args
    • solver: Optimizer type, example: SGD, Adam.
    • others: Please refer to torch.optim.

scheduler_cfg

  • Learning rate scheduler
  • Args
    • scheduler : Learning rate scheduler, example: OneCycleLR.
    • others : Please refer to torch.optim.lr_scheduler.

model_cfg

  • Model to be trained

evaluator_cfg

  • Evaluator configuration
  • Args
    • enable_float16: If true, enable the auto mixed precision mode, default: false.
    • restore_ckpt_strict: If true, check whether the checkpoint is the same as the defined model.
    • restore_hint: int value indicates the iteration number of restored checkpoint, str value indicates the path of restored checkpoint.
    • save_name: The name of the experiment.
    • eval_func: The function name of evaluation.
    • sampler:
      • type: The name of sampler. Choose InferenceSampler.
      • sample_type: In general, we use all_ordered to input all frames by the natural order, which makes sure the tests are consistent.
      • batch_size: int values.
      • others: Please refer to data.sampler and data.collate_fn
    • transform: Support pose data transform. GaitTR_MultiInput, SkeletonInput, GaitGraph_MultiInput
    • metric: euc or cos.

trainer_cfg

  • Trainer configuration
  • Args
    • restore_hint: int value indicates the iteration number of restored checkpoint, str value indicates the path to restored checkpoint. The option is often used to finetune on new dataset or restore the interrupted training process.
    • fix_BN: If True, we fix the weight of all BatchNorm layers.
    • log_iter: Log the information per log_iter iterations.
    • save_iter: Save the checkpoint per save_iter iterations.
    • with_test: If True, we test the model every save_iter iterations. (Disable in Default)
    • optimizer_reset: If True and restore_hint!=0, reset the optimizer while restoring the model.
    • scheduler_reset: If True and restore_hint!=0, reset the scheduler while restoring the model.
    • sync_BN: If True, applies Batch Normalization synchronously.
    • total_iter: The total training iterations, int values.
    • sampler:
      • type: The name of sampler. Support TripletSampler and CommonSampler
      • sample_type: [all, fixed, unfixed] indicates the number of frames used to test, while [unordered, ordered] means whether input sequence by the natural order. Example: fixed_unordered means selecting fixed number of frames randomly.
      • batch_size: [P,K] for TripletSampler where P denotes the subjects in one training batch while K represents the sequences every subject owns. For CommonSampler, [B]
      • others: Please refer to data.sampler and data.collate_fn.
    • others: Please refer to evaluator_cfg.

Note:

  • All the config items will be merged into default.yaml, and the current config is preferable.
  • The output directory, which includes the log, checkpoint and summary files, depends on the defined dataset_name, model and save_name settings, like output/${dataset_name}/${model}/${save_name}.

Example

data_cfg:
  dataset_name: CASIA-B
  dataset_root: your_path
  num_workers: 1
  dataset_partition: ./datasets/CASIA-B/CASIA-B.json
  remove_no_gallery: false
  cache: true
  frame_threshold: 0
  test_dataset_name: CASIA-B

evaluator_cfg:
  enable_float16: false
  restore_ckpt_strict: true
  restore_hint: 80000
  save_name: tmp
  eval_func: identification
  sampler:
    batch_size: 4
    sample_type: all_ordered
    type: InferenceSampler
  transform:
    - type: SkeletonInput

loss_cfg:
  loss_term_weight: 1.0
  margin: 0.2
  type: TripletLoss
  log_prefix: triplet

model_cfg:
  model: tmp

optimizer_cfg:
  lr: 0.1
  momentum: 0.9
  solver: SGD
  weight_decay: 0.0005

scheduler_cfg:
  max_lr: 0.01
  total_steps: 20000
  scheduler: OneCycleLR

trainer_cfg:
  enable_float16: false
  with_test: false
  fix_BN: false
  log_iter: 100
  restore_ckpt_strict: true
  optimizer_reset: false
  scheduler_reset: false
  restore_hint: 0
  save_iter: 2000
  save_name: tmp
  sync_BN: false
  total_iter: 80000
  sampler:
    batch_shuffle: false
    batch_size:
      - 8
      - 16
    frames_num_fixed: 30
    frames_num_max: 50
    frames_num_min: 25
    sample_type: fixed_unordered
    type: TripletSampler
  transform:
    - type: SkeletonInput