train_net_u.py

#!/usr/bin/env python3
# Copyright (c) Facebook, Inc. and its affiliates.

"""
Panoptic-DeepLab Training Script.
This script is a simplified version of the training script in detectron2/tools.
"""

import logging
import os
import time
import weakref
import torch

import detectron2.data.transforms as T
import detectron2.utils.comm as comm
from detectron2.checkpoint import DetectionCheckpointer
from detectron2.config import get_cfg
from detectron2.data import MetadataCatalog, build_detection_train_loader
from detectron2.engine import (
    DefaultTrainer,
    SimpleTrainer,
    AMPTrainer,
    default_argument_parser,
    default_setup,
    launch,
    create_ddp_model,
)
from detectron2.evaluation import (
    CityscapesInstanceEvaluator,
    CityscapesSemSegEvaluator,
    COCOEvaluator,
    COCOPanopticEvaluator,
    DatasetEvaluators,
)
from detectron2.utils.logger import _log_api_usage
from detectron2.projects.deeplab import build_lr_scheduler
from detectron2.projects.panoptic_deeplab import (
    PanopticDeeplabDatasetMapper,
    add_panoptic_deeplab_config,
)
from detectron2.solver import get_default_optimizer_params
from detectron2.solver.build import maybe_add_gradient_clipping

# Custom
import detectron2.projects.panoptic_deeplab.datasets
from detectron2.projects.panoptic_deeplab import InstCalU


def set_all_requires_grad_false(module):
    for module in module.modules():
        for param in module.parameters():
            param.requires_grad = False


def set_norm_requires_grad(module, flag=True):
    for module in module.modules():
        for name, param in module.named_parameters():
            if 'momentum' in name:
                param.requires_grad = flag


def build_sem_seg_train_aug(cfg):
    augs = [
        T.ResizeShortestEdge(
            cfg.INPUT.MIN_SIZE_TRAIN, cfg.INPUT.MAX_SIZE_TRAIN, cfg.INPUT.MIN_SIZE_TRAIN_SAMPLING
        )
    ]
    if cfg.INPUT.CROP.ENABLED:
        augs.append(T.RandomCrop(cfg.INPUT.CROP.TYPE, cfg.INPUT.CROP.SIZE))
    augs.append(T.RandomFlip())
    return augs


class Trainer(DefaultTrainer):
    """
    We use the "DefaultTrainer" which contains a number pre-defined logic for
    standard training workflow. They may not work for you, especially if you
    are working on a new research project. In that case you can use the cleaner
    "SimpleTrainer", or write your own training loop.
    """

    def __init__(self, cfg):
        """
        Args:
            cfg (CfgNode):
        """
        # From TrainerBase
        self._hooks = []
        self.iter = 0
        self.start_iter = 0
        assert not cfg.SOLVER.AMP.ENABLED, 'Not implemented'

        logger = logging.getLogger("detectron2")
        if not logger.isEnabledFor(logging.INFO):  # setup_logger is not called for d2
            setup_logger()
        cfg = DefaultTrainer.auto_scale_workers(cfg, comm.get_world_size())

        # Assume these objects must be constructed in this order.
        model = self.build_model(cfg)
        data_loader = self.build_train_loader(cfg)
        model = create_ddp_model(model, broadcast_buffers=False)
        self.checkpointer = DetectionCheckpointer(
            # Assume you want to save checkpoints together with logs/statistics
            model,
            cfg.OUTPUT_DIR,
            trainer=weakref.proxy(self),
        )

        self.start_iter = 0
        self.max_iter = cfg.SOLVER.MAX_ITER
        self.cfg = cfg
        self.register_hooks(self.build_hooks())

        self.resume_or_load(resume=args.resume)
        self.checkpointer.model = InstCalU.convert_adaptive_batchnorm(self.checkpointer.model, per_channel=True, separate=True, init=0.1)
        self.checkpointer.model.cuda()
        set_all_requires_grad_false(self.checkpointer.model)
        set_norm_requires_grad(self.checkpointer.model, True)

        optimizer = self.build_optimizer(cfg, model)
        self._trainer = (AMPTrainer if cfg.SOLVER.AMP.ENABLED else SimpleTrainer)(
            model, data_loader, optimizer
        )
        self.scheduler = self.build_lr_scheduler(cfg, optimizer)

    @classmethod
    def build_evaluator(cls, cfg, dataset_name, output_folder=None):
        """
        Create evaluator(s) for a given dataset.
        This uses the special metadata "evaluator_type" associated with each builtin dataset.
        For your own dataset, you can simply create an evaluator manually in your
        script and do not have to worry about the hacky if-else logic here.
        """
        if cfg.MODEL.PANOPTIC_DEEPLAB.BENCHMARK_NETWORK_SPEED:
            return None
        if output_folder is None:
            output_folder = os.path.join(cfg.OUTPUT_DIR, "inference")
        evaluator_list = []
        evaluator_type = MetadataCatalog.get(dataset_name).evaluator_type
        if evaluator_type in ["cityscapes_panoptic_seg", "coco_panoptic_seg"]:
            evaluator_list.append(COCOPanopticEvaluator(dataset_name, output_folder))
        if evaluator_type == "cityscapes_panoptic_seg":
            evaluator_list.append(CityscapesSemSegEvaluator(dataset_name))
            evaluator_list.append(CityscapesInstanceEvaluator(dataset_name))
        if evaluator_type == "coco_panoptic_seg":
            # `thing_classes` in COCO panoptic metadata includes both thing and
            # stuff classes for visualization. COCOEvaluator requires metadata
            # which only contains thing classes, thus we map the name of
            # panoptic datasets to their corresponding instance datasets.
            dataset_name_mapper = {
                "coco_2017_val_panoptic": "coco_2017_val",
                "coco_2017_val_100_panoptic": "coco_2017_val_100",
            }
            evaluator_list.append(
                COCOEvaluator(dataset_name_mapper[dataset_name], output_dir=output_folder)
            )
        if len(evaluator_list) == 0:
            raise NotImplementedError(
                "no Evaluator for the dataset {} with the type {}".format(
                    dataset_name, evaluator_type
                )
            )
        elif len(evaluator_list) == 1:
            return evaluator_list[0]
        return DatasetEvaluators(evaluator_list)

    @classmethod
    def build_train_loader(cls, cfg):
        mapper = PanopticDeeplabDatasetMapper(cfg, augmentations=build_sem_seg_train_aug(cfg))
        return build_detection_train_loader(cfg, mapper=mapper)

    @classmethod
    def build_lr_scheduler(cls, cfg, optimizer):
        """
        It now calls :func:`detectron2.solver.build_lr_scheduler`.
        Overwrite it if you'd like a different scheduler.
        """
        return build_lr_scheduler(cfg, optimizer)

    @classmethod
    def build_optimizer(cls, cfg, model):
        """
        Build an optimizer from config.
        """
        params = get_default_optimizer_params(
            model,
            weight_decay=cfg.SOLVER.WEIGHT_DECAY,
            weight_decay_norm=cfg.SOLVER.WEIGHT_DECAY_NORM,
        )

        optimizer_type = cfg.SOLVER.OPTIMIZER
        if optimizer_type == "SGD":
            return maybe_add_gradient_clipping(cfg, torch.optim.SGD)(
                params,
                cfg.SOLVER.BASE_LR,
                momentum=cfg.SOLVER.MOMENTUM,
                nesterov=cfg.SOLVER.NESTEROV,
            )
        elif optimizer_type == "ADAM":
            return maybe_add_gradient_clipping(cfg, torch.optim.Adam)(params, cfg.SOLVER.BASE_LR)
        else:
            raise NotImplementedError(f"no optimizer type {optimizer_type}")


def setup(args):
    """
    Create configs and perform basic setups.
    """
    cfg = get_cfg()
    add_panoptic_deeplab_config(cfg)
    cfg.merge_from_file(args.config_file)
    cfg.merge_from_list(args.opts)
    cfg.freeze()
    default_setup(cfg, args)
    return cfg


def main(args):
    cfg = setup(args)

    if args.eval_only:
        model = Trainer.build_model(cfg)

        # NOTE: Need to convert first to load weights correctly
        model = InstCalU.convert_adaptive_batchnorm(model, per_channel=True, separate=True, init=0.1)
        model.cuda()
        DetectionCheckpointer(model, save_dir=cfg.OUTPUT_DIR).resume_or_load(
            cfg.MODEL.WEIGHTS, resume=args.resume
        )

        res = Trainer.test(cfg, model)
        return res

    trainer = Trainer(cfg)
    #trainer.resume_or_load(resume=args.resume)
    return trainer.train()


if __name__ == "__main__":
    args = default_argument_parser().parse_args()
    print("Command Line Args:", args)
    launch(
        main,
        args.num_gpus,
        num_machines=args.num_machines,
        machine_rank=args.machine_rank,
        dist_url=args.dist_url,
        args=(args,),
    )