train_net.py

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

from detectron2.checkpoint import DetectionCheckpointer
from detectron2.config import get_cfg
from detectron2.engine import default_argument_parser, default_setup, launch

# hacky way to register
from ubteacher.modeling import *
from ubteacher.engine import *
from ubteacher import add_ubteacher_config
import wandb

from detectron2.data.datasets import register_coco_instances
register_coco_instances("nightowls_train", {}, "/lwll/external/nightowls/annotations/instances_train.json", "/lwll/external/nightowls/train")
register_coco_instances("nightowls_val", {}, "/lwll/external/nightowls/annotations/instances_val.json", "/lwll/external/nightowls/val")


def setup(args):
    """
    Create configs and perform basic setups.
    """
    cfg = get_cfg()

    add_ubteacher_config(cfg)
    cfg.merge_from_file(args.config_file)
    cfg.merge_from_list(args.opts)
    cfg.freeze()
    default_setup(cfg, args)
    wandb.init(project='nightowls', sync_tensorboard=True,
           settings=wandb.Settings(start_method="thread", console="off"), config=cfg, )
    wandb.run.name = cfg.EXPERIMENT_NAME
    return cfg


def main(args):
    cfg = setup(args)
    if cfg.SEMISUPNET.Trainer == "ubteacher":
        Trainer = UBTeacherTrainer
    elif cfg.SEMISUPNET.Trainer == "ubteacher_rcnn":
        Trainer = UBRCNNTeacherTrainer
    else:
        raise ValueError("Trainer Name is not found.")

    if args.eval_only:
        if cfg.SEMISUPNET.Trainer == "ubteacher":
            model = Trainer.build_model(cfg)
            model_teacher = Trainer.build_model(cfg)
            ensem_ts_model = EnsembleTSModel(model_teacher, model)

            DetectionCheckpointer(
                ensem_ts_model, save_dir=cfg.OUTPUT_DIR
            ).resume_or_load(cfg.MODEL.WEIGHTS, resume=args.resume)
            res = Trainer.test(cfg, ensem_ts_model.modelTeacher)

        else:
            model = Trainer.build_model(cfg)
            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,),
    )