-
Notifications
You must be signed in to change notification settings - Fork 681
/
compute_metric.py
147 lines (121 loc) · 5.78 KB
/
compute_metric.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
# Copyright (c) MONAI Consortium
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
# http://www.apache.org/licenses/LICENSE-2.0
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""
This example shows how to efficiently compute Dice scores for pairs of segmentation prediction
and references in multi-processing based on MONAI's metrics API.
It can even run on multi-nodes.
Main steps to set up the distributed data parallel:
- Execute `torchrun` to create processes on every node for every process.
It receives parameters as below:
`--nproc_per_node=NUM_PROCESSES_PER_NODE`
`--nnodes=NUM_NODES`
For more details, refer to https://github.com/pytorch/pytorch/blob/master/torch/distributed/run.py.
Alternatively, we can also use `torch.multiprocessing.spawn` to start program, but it that case, need to handle
all the above parameters and compute `rank` manually, then set to `init_process_group`, etc.
`torchrun` is even more efficient than `torch.multiprocessing.spawn`.
- Use `init_process_group` to initialize every process.
- Partition the saved predictions and labels into ranks for parallel computation.
- Compute `Dice Metric` on every process, reduce the results after synchronization.
Note:
`torchrun` will launch `nnodes * nproc_per_node = world_size` processes in total.
Example script to execute this program on a single node with 2 processes:
`torchrun --nproc_per_node=2 compute_metric.py`
Referring to: https://pytorch.org/tutorials/intermediate/ddp_tutorial.html
"""
import argparse
import os
from glob import glob
import nibabel as nib
import numpy as np
import torch
import torch.distributed as dist
from monai.data import create_test_image_3d, partition_dataset
from monai.handlers import write_metrics_reports
from monai.metrics import DiceMetric
from monai.transforms import (
AsDiscreted,
EnsureChannelFirstd,
Compose,
KeepLargestConnectedComponentd,
LoadImaged,
ScaleIntensityd,
ToDeviced,
)
from monai.utils import string_list_all_gather
def compute(args):
# generate synthetic data for the example
local_rank = int(os.environ["LOCAL_RANK"])
if local_rank == 0 and not os.path.exists(args.dir):
# create 16 random pred, label paris for evaluation
print(f"generating synthetic data to {args.dir} (this may take a while)")
os.makedirs(args.dir)
# if have multiple nodes, set random seed to generate same random data for every node
np.random.seed(seed=0)
for i in range(16):
pred, label = create_test_image_3d(128, 128, 128, num_seg_classes=1, channel_dim=-1, noise_max=0.5)
n = nib.Nifti1Image(pred, np.eye(4))
nib.save(n, os.path.join(args.dir, f"pred{i:d}.nii.gz"))
n = nib.Nifti1Image(label, np.eye(4))
nib.save(n, os.path.join(args.dir, f"label{i:d}.nii.gz"))
# initialize the distributed evaluation process, change to gloo backend if computing on CPU
dist.init_process_group(backend="nccl", init_method="env://")
preds = sorted(glob(os.path.join(args.dir, "pred*.nii.gz")))
labels = sorted(glob(os.path.join(args.dir, "label*.nii.gz")))
datalist = [{"pred": pred, "label": label} for pred, label in zip(preds, labels)]
# split data for every subprocess, for example, 16 processes compute in parallel
data_part = partition_dataset(
data=datalist,
num_partitions=dist.get_world_size(),
shuffle=False,
even_divisible=False,
)[dist.get_rank()]
device = torch.device(f"cuda:{local_rank}")
torch.cuda.set_device(device)
# define transforms for predictions and labels
transforms = Compose(
[
LoadImaged(keys=["pred", "label"]),
ToDeviced(keys=["pred", "label"], device=device),
EnsureChannelFirstd(keys=["pred", "label"]),
ScaleIntensityd(keys="pred"),
AsDiscreted(keys="pred", threshold=0.5),
KeepLargestConnectedComponentd(keys="pred", applied_labels=[1]),
]
)
data_part = [transforms(item) for item in data_part]
# compute metrics for current process
metric = DiceMetric(include_background=True, reduction="mean", get_not_nans=False)
metric(y_pred=[i["pred"] for i in data_part], y=[i["label"] for i in data_part])
filenames = [item["pred_meta_dict"]["filename_or_obj"] for item in data_part]
# all-gather results from all the processes and reduce for final result
result = metric.aggregate().item()
filenames = string_list_all_gather(strings=filenames)
if local_rank == 0:
print("mean dice: ", result)
# generate metrics reports at: output/mean_dice_raw.csv, output/mean_dice_summary.csv, output/metrics.csv
write_metrics_reports(
save_dir="./output",
images=filenames,
metrics={"mean_dice": result},
metric_details={"mean_dice": metric.get_buffer()},
summary_ops="*",
)
metric.reset()
dist.destroy_process_group()
# usage example(refer to https://github.com/pytorch/pytorch/blob/master/torch/distributed/launch.py):
# torchrun --standalone --nnodes=NUM_NODES --nproc_per_node=NUM_GPUS_PER_NODE compute_metric.py -d DIR_OF_OUTPUT
def main():
parser = argparse.ArgumentParser()
parser.add_argument("-d", "--dir", default="./output", type=str, help="root directory of labels and predictions.")
args = parser.parse_args()
compute(args=args)
if __name__ == "__main__":
main()