{
    "imports": [
        "$import glob",
        "$import os"
    ],
    "bundle_root": "./",
    "ckpt_dir": "$@bundle_root + '/models'",
    "output_dir": "$@bundle_root + '/eval'",
    "data_list_file_path": "$@bundle_root + '/annotation/dataset_fold0.json'",
    "data_file_base_dir": "/home/canz/Projects/datasets/LUNA16/93176/Images_resample",
    "train_datalist": "$monai.data.load_decathlon_datalist(@data_list_file_path, is_segmentation=True, data_list_key='training', base_dir=@data_file_base_dir)",
    "device": "$torch.device('cuda:0' if torch.cuda.is_available() else 'cpu')",
    "epochs": 300,
    "val_interval": 10,
    "learning_rate": 0.01,
    "amp": true,
    "batch_size": 3,
    "patch_size": [
        192,
        192,
        80
    ],
    "val_patch_size": [
        512,
        512,
        208
    ],
    "anchor_generator": {
        "_target_": "monai.apps.detection.utils.anchor_utils.AnchorGeneratorWithAnchorShape",
        "feature_map_scales": [
            1,
            2,
            4
        ],
        "base_anchor_shapes": [
            [
                6,
                8,
                4
            ],
            [
                8,
                6,
                5
            ],
            [
                10,
                10,
                6
            ]
        ]
    },
    "backbone": "$monai.networks.nets.resnet.resnet50(spatial_dims=3,n_input_channels=1,conv1_t_stride=[2,2,1],conv1_t_size=[7,7,7])",
    "feature_extractor": "$monai.apps.detection.networks.retinanet_network.resnet_fpn_feature_extractor(@backbone,3,False,[1,2],None)",
    "network_def": {
        "_target_": "RetinaNet",
        "spatial_dims": 3,
        "num_classes": 1,
        "num_anchors": 3,
        "feature_extractor": "@feature_extractor",
        "size_divisible": [
            16,
            16,
            8
        ]
    },
    "network": "$@network_def.to(@device)",
    "detector": {
        "_target_": "RetinaNetDetector",
        "network": "@network",
        "anchor_generator": "@anchor_generator",
        "debug": false
    },
    "detector_ops": [
        "$@detector.set_atss_matcher(num_candidates=4, center_in_gt=False)",
        "$@detector.set_hard_negative_sampler(batch_size_per_image=64,positive_fraction=0.3,pool_size=20,min_neg=16)",
        "$@detector.set_target_keys(box_key='box', label_key='label')",
        "$@detector.set_box_selector_parameters(score_thresh=0.02,topk_candidates_per_level=1000,nms_thresh=0.22,detections_per_img=300)",
        "$@detector.set_sliding_window_inferer(roi_size=@val_patch_size,overlap=0.25,sw_batch_size=1,mode='constant',device='cpu')"
    ],
    "optimizer": {
        "_target_": "torch.optim.SGD",
        "params": "$@detector.network.parameters()",
        "lr": "@learning_rate",
        "momentum": 0.9,
        "weight_decay": 3e-05,
        "nesterov": true
    },
    "after_scheduler": {
        "_target_": "torch.optim.lr_scheduler.StepLR",
        "optimizer": "@optimizer",
        "step_size": 150,
        "gamma": 0.1
    },
    "lr_scheduler": {
        "_target_": "scripts.warmup_scheduler.GradualWarmupScheduler",
        "optimizer": "@optimizer",
        "multiplier": 1,
        "total_epoch": 10,
        "after_scheduler": "@after_scheduler"
    },
    "train": {
        "preprocessing_transforms": [
            {
                "_target_": "LoadImaged",
                "keys": "image",
                "meta_key_postfix": "meta_dict"
            },
            {
                "_target_": "EnsureChannelFirstd",
                "keys": "image",
                "meta_key_postfix": "meta_dict"
            },
            {
                "_target_": "EnsureTyped",
                "keys": [
                    "image",
                    "box"
                ]
            },
            {
                "_target_": "EnsureTyped",
                "keys": "label",
                "dtype": "$torch.long"
            },
            {
                "_target_": "Orientationd",
                "keys": "image",
                "axcodes": "RAS"
            },
            {
                "_target_": "ScaleIntensityRanged",
                "keys": "image",
                "a_min": -1024.0,
                "a_max": 300.0,
                "b_min": 0.0,
                "b_max": 1.0,
                "clip": true
            },
            {
                "_target_": "ConvertBoxToStandardModed",
                "box_keys": "box",
                "mode": "cccwhd"
            },
            {
                "_target_": "AffineBoxToImageCoordinated",
                "box_keys": "box",
                "box_ref_image_keys": "image",
                "image_meta_key_postfix": "meta_dict",
                "affine_lps_to_ras": true
            }
        ],
        "random_transforms": [
            {
                "_target_": "RandCropBoxByPosNegLabeld",
                "image_keys": "image",
                "box_keys": "box",
                "label_keys": "label",
                "spatial_size": "@patch_size",
                "whole_box": true,
                "num_samples": "@batch_size",
                "pos": 1,
                "neg": 1
            },
            {
                "_target_": "RandZoomBoxd",
                "image_keys": "image",
                "box_keys": "box",
                "label_keys": "label",
                "box_ref_image_keys": "image",
                "prob": 0.2,
                "min_zoom": 0.7,
                "max_zoom": 1.4,
                "padding_mode": "constant",
                "keep_size": true
            },
            {
                "_target_": "ClipBoxToImaged",
                "box_keys": "box",
                "label_keys": "label",
                "box_ref_image_keys": "image",
                "remove_empty": true
            },
            {
                "_target_": "RandFlipBoxd",
                "image_keys": "image",
                "box_keys": "box",
                "box_ref_image_keys": "image",
                "prob": 0.5,
                "spatial_axis": 0
            },
            {
                "_target_": "RandFlipBoxd",
                "image_keys": "image",
                "box_keys": "box",
                "box_ref_image_keys": "image",
                "prob": 0.5,
                "spatial_axis": 1
            },
            {
                "_target_": "RandFlipBoxd",
                "image_keys": "image",
                "box_keys": "box",
                "box_ref_image_keys": "image",
                "prob": 0.5,
                "spatial_axis": 2
            },
            {
                "_target_": "RandRotateBox90d",
                "image_keys": "image",
                "box_keys": "box",
                "box_ref_image_keys": "image",
                "prob": 0.75,
                "max_k": 3,
                "spatial_axes": [
                    0,
                    1
                ]
            },
            {
                "_target_": "BoxToMaskd",
                "box_keys": "box",
                "label_keys": "label",
                "box_mask_keys": "box_mask",
                "box_ref_image_keys": "image",
                "min_fg_label": 0,
                "ellipse_mask": true
            },
            {
                "_target_": "RandRotated",
                "keys": [
                    "image",
                    "box_mask"
                ],
                "mode": [
                    "nearest",
                    "nearest"
                ],
                "prob": 0.2,
                "range_x": 0.5236,
                "range_y": 0.5236,
                "range_z": 0.5236,
                "keep_size": true,
                "padding_mode": "zeros"
            },
            {
                "_target_": "MaskToBoxd",
                "box_keys": [
                    "box"
                ],
                "label_keys": [
                    "label"
                ],
                "box_mask_keys": [
                    "box_mask"
                ],
                "min_fg_label": 0
            },
            {
                "_target_": "DeleteItemsd",
                "keys": "box_mask"
            },
            {
                "_target_": "RandGaussianNoised",
                "keys": "image",
                "prob": 0.1,
                "mean": 0.0,
                "std": 0.1
            },
            {
                "_target_": "RandGaussianSmoothd",
                "keys": "image",
                "prob": 0.1,
                "sigma_x": [
                    0.5,
                    1.0
                ],
                "sigma_y": [
                    0.5,
                    1.0
                ],
                "sigma_z": [
                    0.5,
                    1.0
                ]
            },
            {
                "_target_": "RandScaleIntensityd",
                "keys": "image",
                "factors": 0.25,
                "prob": 0.15
            },
            {
                "_target_": "RandShiftIntensityd",
                "keys": "image",
                "offsets": 0.1,
                "prob": 0.15
            },
            {
                "_target_": "RandAdjustContrastd",
                "keys": "image",
                "prob": 0.3,
                "gamma": [
                    0.7,
                    1.5
                ]
            }
        ],
        "final_transforms": [
            {
                "_target_": "EnsureTyped",
                "keys": [
                    "image",
                    "box"
                ]
            },
            {
                "_target_": "EnsureTyped",
                "keys": "label",
                "dtype": "$torch.long"
            },
            {
                "_target_": "ToTensord",
                "keys": [
                    "image",
                    "box",
                    "label"
                ]
            }
        ],
        "preprocessing": {
            "_target_": "Compose",
            "transforms": "$@train#preprocessing_transforms + @train#random_transforms + @train#final_transforms"
        },
        "dataset": {
            "_target_": "Dataset",
            "data": "$@train_datalist[: int(0.95 * len(@train_datalist))]",
            "transform": "@train#preprocessing"
        },
        "dataloader": {
            "_target_": "DataLoader",
            "dataset": "@train#dataset",
            "batch_size": 1,
            "shuffle": true,
            "num_workers": 4,
            "collate_fn": "$monai.data.utils.no_collation"
        },
        "handlers": [
            {
                "_target_": "LrScheduleHandler",
                "lr_scheduler": "@lr_scheduler",
                "print_lr": true
            },
            {
                "_target_": "ValidationHandler",
                "validator": "@validate#evaluator",
                "epoch_level": true,
                "interval": "@val_interval"
            },
            {
                "_target_": "StatsHandler",
                "tag_name": "train_loss",
                "output_transform": "$lambda x: monai.handlers.from_engine(['loss'], first=True)(x)[0]"
            },
            {
                "_target_": "TensorBoardStatsHandler",
                "log_dir": "@output_dir",
                "tag_name": "train_loss",
                "output_transform": "$lambda x: monai.handlers.from_engine(['loss'], first=True)(x)[0]"
            }
        ],
        "trainer": {
            "_target_": "scripts.trainer.DetectionTrainer",
            "_requires_": "@detector_ops",
            "max_epochs": "@epochs",
            "device": "@device",
            "train_data_loader": "@train#dataloader",
            "detector": "@detector",
            "optimizer": "@optimizer",
            "train_handlers": "@train#handlers",
            "amp": "@amp"
        }
    },
    "validate": {
        "preprocessing": {
            "_target_": "Compose",
            "transforms": "$@train#preprocessing_transforms + @train#final_transforms"
        },
        "dataset": {
            "_target_": "Dataset",
            "data": "$@train_datalist[int(0.95 * len(@train_datalist)): ]",
            "transform": "@validate#preprocessing"
        },
        "dataloader": {
            "_target_": "DataLoader",
            "dataset": "@validate#dataset",
            "batch_size": 1,
            "shuffle": false,
            "num_workers": 2,
            "collate_fn": "$monai.data.utils.no_collation"
        },
        "handlers": [
            {
                "_target_": "StatsHandler",
                "iteration_log": false
            },
            {
                "_target_": "TensorBoardStatsHandler",
                "log_dir": "@output_dir",
                "iteration_log": false
            },
            {
                "_target_": "CheckpointSaver",
                "save_dir": "@ckpt_dir",
                "save_dict": {
                    "model": "@network"
                },
                "save_key_metric": true,
                "key_metric_filename": "model.pt"
            }
        ],
        "key_metric": {
            "val_coco": {
                "_target_": "scripts.cocometric_ignite.IgniteCocoMetric",
                "coco_metric_monai": "$monai.apps.detection.metrics.coco.COCOMetric(classes=['nodule'], iou_list=[0.1], max_detection=[100])",
                "output_transform": "$monai.handlers.from_engine(['pred', 'label'])",
                "box_key": "box",
                "label_key": "label",
                "pred_score_key": "label_scores",
                "reduce_scalar": true
            }
        },
        "evaluator": {
            "_target_": "scripts.evaluator.DetectionEvaluator",
            "_requires_": "@detector_ops",
            "device": "@device",
            "val_data_loader": "@validate#dataloader",
            "detector": "@detector",
            "key_val_metric": "@validate#key_metric",
            "val_handlers": "@validate#handlers",
            "amp": "@amp"
        }
    },
    "training": [
        "os.environ['CUDA_LAUNCH_BLOCKING']=1",
        "$monai.utils.set_determinism(seed=123)",
        "$setattr(torch.backends.cudnn, 'benchmark', True)",
        "$@train#trainer.run()"
    ]
}