mmcls调用逻辑

dataloader

IterBaseRunner

    + register_training_hooks()

    + run()    + train()    + val()    + resume()    + save_checkpoint()

optimizer

配置参数config（dict）model，dataset，schedules，runtime

utils/register.py

openmmlab-cls

装饰

OPTIMIZER_BUILDERS = Registry('optimizer builder')

DATASETS = Registry('dataset')PIPELINES = Registry('pipeline')SAMPLERS = Registry('sampler')

model

构建modelbuild_classifier()

runner

Runner

Config

+ fromfile(filename):return cfg

EpochBaseRunner

    + run()    + run_iter()    + train()

    + val()    + save_checkpoint()

dataset

datasets

构建datasetsbuild_dataset()

runner/optimizer/build.py

dataloader=build_dataloader(dataset)

Hook

+ before_run(runner)+ before_train_epoch(runner)+ before_train_iter(runner)+ after_train_iter(runner)+ after_train_epoch(runner)+ before_val_epoch(runner)+ before_val_iter(runner)+ after_val_iter(runner)+ after_val_epoch(runner)+ after_run(runner)+ get_triggered_stages()

runner/hooks

cnn

CustomHook

+ custom_function(runner)

train.py

runner.register_hook()

RUNNER_BUILDERS.build(cfg)

cfg=Config().fromfile()

cnn.bricks.registry

OPTIMIZERS = Registry('optimizer')

RUNNER = Registry('runner')

runner.register_training_hooks()

train_model流程

MODELS

+ build_classifier(cfg)+ build_backbone(cfg)+ build_head(cfg)+ build_neck(cfg)+ build_loss(cfg)

EvalHook

+ evaluation()

+ before_run(runner)+ before_train_epoch(runner)+ before_train_iter(runner)+ after_train_iter(runner)+ after_train_epoch(runner)

runner/build.py

build_from_cfg()

OptimizerHook

+ after_train_iter(runner)

mmcv

dataset=build_dataset(cfg.data)

CheckpointHook

+ before_run(runner)

+ after_train_iter(runner)+ after_train_epoch(runner)

config

RUNNER_BUILDERS = Registry('runner builder')

Register

+ build() # 编译模块（接口）+ build_func() # 编译（外部函数）

+ register_module() # 注册模块

LrUpdaterHook

+ get_lr(runner)+ get_regular_lr(runner)+ get_warmup_lr(runner)

+ before_run(runner)+ after_train_iter(runner)+ after_train_epoch(runner)

mmcls

BaseRunner

+ run()# 虚函数，在继承类中实现+ train()# 虚函数，在继承类中实现+ val()# 虚函数，在继承类中实现+ save_checkpoint()

- _hooks+ register_hook():return _hooks+ call_hook()+ resume()+ register_lr_hook(cfg)+ register_momentum_hook(cfg)+ register_optimizer_hook(cfg)+ register_checkpoint_hook(cfg)+ register_logger_hook(cfg)+ register_timer_hook(cfg)+ register_custom_hook(cfg)+ register_training_hook(cfg)

HOOKS = Registry('hook')

tools

MMCV_ATTENTION=Register('attention')

logger

api/train.py