forked from IsHYuhi/FujiFilm_Brains_Solution
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Q2_classification_inference.py
143 lines (112 loc) · 4.24 KB
/
Q2_classification_inference.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
import argparse
import os
from typing import List, Union
import numpy as np
import pandas as pd
import timm
import torch
import torch.nn as nn
from libs.config import Config, get_config
from libs.data_loader import ImageDataset, ImageTransform, make_datapath_list
from libs.models import fix_model_state_dict, get_fcn
from train import calc_acc_f1, seed_everything
seed_everything(seed=42)
def get_parser() -> argparse.Namespace:
parser = argparse.ArgumentParser(
prog="semantic classification inference",
usage="python3 Q2_inference.py",
description="""
This module demonstrates classification inference.
""",
add_help=True,
)
parser.add_argument("config", type=str, help="path of a config file")
parser.add_argument(
"--top-k", type=int, default=5, help="top-k of validation score"
)
return parser.parse_args()
def calc_ensemble(preds: np.ndarray, th: int) -> np.ndarray:
ens_preds = np.zeros_like(preds[0])
for pred in preds:
ens_preds += pred
ens_preds = np.where(ens_preds >= th, 1, 0)
return ens_preds
def predict(net: nn.Module, dataset: ImageDataset, device: str) -> np.ndarray:
net.eval()
preds = []
for i in range(dataset.__len__()):
img = dataset[i]
_, h, w = img.shape
img = torch.unsqueeze(img, dim=0)
with torch.no_grad():
pred = net(img.to(device))
_, pred = torch.max(pred, dim=1)
preds.append(pred.squeeze().detach().cpu().numpy())
preds = np.array(preds)
return preds
def save_csv(preds: Union[List, np.ndarray], img_list: List, save_name: str) -> None:
submit = pd.DataFrame()
submit["img"] = img_list
submit["label"] = list(preds)
submit.to_csv("submission_csv/{:s}.csv".format(save_name), header=None, index=None)
def main(parser: Config, load_model_name: str, top_k: int) -> None:
if not os.path.exists("./submission_csv"):
os.mkdir("./submission_csv")
ens = []
preds = []
mean = (0.5,)
std = (0.5,)
device = "cuda" if torch.cuda.is_available() else "cpu"
test_img_list = make_datapath_list(phase="test", dataset_name=parser.dataset_name)
test_dataset = ImageDataset(
img_list=test_img_list,
img_transform=ImageTransform(size=parser.image_size, mean=mean, std=std),
phase="test",
)
for fold_id in range(0, 5):
print("fold_{:d}".format(fold_id))
if parser.model_name == "fcn":
net = get_fcn(
pretrained=False, in_channels=3, out_channels=parser.num_classes
)
else:
net = timm.create_model(
parser.model_name, pretrained=False, num_classes=parser.num_classes
)
_, val_img_list = make_datapath_list(
phase="train", n_splits=5, fold_id=fold_id, dataset_name=parser.dataset_name
)
val_dataset = ImageDataset(
img_list=val_img_list,
img_transform=ImageTransform(size=parser.image_size, mean=mean, std=std),
phase="val",
)
checkpoint_model_name = "{:s}/fold{:d}".format(
load_model_name,
fold_id,
)
save_csv_name = "{:s}_fold{:d}".format(load_model_name, fold_id)
net_weights = torch.load(
"./checkpoints/" + checkpoint_model_name + "_max_val_f1_net.pth",
map_location=torch.device(device),
)
net.load_state_dict(fix_model_state_dict(net_weights))
net.to(device)
net.eval()
val_acc, val_f1 = calc_acc_f1(net, val_dataset, device)
print(
"validation accuracy: {:f} || validation f1 score: {:f}".format(
val_acc, val_f1
)
)
test_preds = predict(net, test_dataset, device)
save_csv(test_preds, test_img_list["img"], save_csv_name)
preds.append([test_preds, val_f1])
preds = sorted(preds, key=lambda x: x[1], reverse=True)
ens = calc_ensemble([x[0] for x in preds[:top_k]], top_k)
save_csv(ens, test_img_list["img"], load_model_name)
if __name__ == "__main__":
parser = get_parser()
save_name = parser.config.split("/")[-1].split(".")[0]
config = get_config(parser.config)
main(config, save_name, parser.top_k)