configs/inference.json

{
    "imports": [
        "$import glob",
        "$import numpy",
        "$import os"
    ],
    "bundle_root": ".",
    "image_key": "image",
    "output_dir": "$@bundle_root + '/eval'",
    "output_ext": ".png",
    "output_dtype": "$numpy.float32",
    "output_postfix": "trans",
    "separate_folder": true,
    "load_pretrain": true,
    "dataset_dir": "/workspace/data/endoscopic_tool_dataset",
    "datalist": "$list(sorted(glob.glob(os.path.join(@dataset_dir,'test', '*', '*[!seg].jpg'))))",
    "device": "$torch.device('cuda:0' if torch.cuda.is_available() else 'cpu')",
    "network_def": {
        "_target_": "FlexibleUNet",
        "in_channels": 3,
        "out_channels": 2,
        "backbone": "efficientnet-b2",
        "spatial_dims": 2,
        "pretrained": false,
        "is_pad": false,
        "pre_conv": null
    },
    "network": "$@network_def.to(@device)",
    "preprocessing": {
        "_target_": "Compose",
        "transforms": [
            {
                "_target_": "LoadImaged",
                "keys": "@image_key",
                "image_only": false
            },
            {
                "_target_": "EnsureChannelFirstd",
                "keys": "@image_key",
                "channel_dim": -1
            },
            {
                "_target_": "Resized",
                "keys": [
                    "@image_key"
                ],
                "spatial_size": [
                    736,
                    480
                ],
                "mode": [
                    "bilinear"
                ]
            },
            {
                "_target_": "ScaleIntensityd",
                "keys": [
                    "@image_key"
                ]
            }
        ]
    },
    "dataset": {
        "_target_": "Dataset",
        "data": "$[{'image': i} for i in @datalist]",
        "transform": "@preprocessing"
    },
    "dataloader": {
        "_target_": "DataLoader",
        "dataset": "@dataset",
        "batch_size": 1,
        "shuffle": false,
        "num_workers": 4
    },
    "inferer": {
        "_target_": "SimpleInferer"
    },
    "postprocessing": {
        "_target_": "Compose",
        "transforms": [
            {
                "_target_": "Invertd",
                "keys": "pred",
                "transform": "@preprocessing",
                "orig_keys": "@image_key",
                "meta_key_postfix": "meta_dict",
                "nearest_interp": false,
                "to_tensor": true
            },
            {
                "_target_": "AsDiscreted",
                "argmax": true,
                "to_onehot": 2,
                "keys": [
                    "pred"
                ]
            },
            {
                "_target_": "Lambdad",
                "keys": [
                    "pred"
                ],
                "func": "$lambda x : x[1:]"
            },
            {
                "_target_": "SaveImaged",
                "keys": "pred",
                "meta_keys": "pred_meta_dict",
                "output_dir": "@output_dir",
                "output_ext": "@output_ext",
                "output_dtype": "@output_dtype",
                "output_postfix": "@output_postfix",
                "separate_folder": "@separate_folder",
                "scale": 255,
                "squeeze_end_dims": true
            }
        ]
    },
    "handlers": [
        {
            "_target_": "StatsHandler",
            "iteration_log": false
        }
    ],
    "evaluator": {
        "_target_": "SupervisedEvaluator",
        "device": "@device",
        "val_data_loader": "@dataloader",
        "network": "@network",
        "inferer": "@inferer",
        "postprocessing": "@postprocessing",
        "val_handlers": "@handlers"
    },
    "checkpointloader": {
        "_target_": "CheckpointLoader",
        "load_path": "$@bundle_root + '/models/model.pt'",
        "load_dict": {
            "model": "@network"
        },
        "map_location": "@device"
    },
    "initialize": [
        "$monai.utils.set_determinism(seed=123)",
        "$@checkpointloader(@evaluator) if @load_pretrain else None"
    ],
    "run": [
        "[email protected]()"
    ]
}