-
Notifications
You must be signed in to change notification settings - Fork 18
/
evaluation.py
140 lines (117 loc) · 6.56 KB
/
evaluation.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
import os
import json
import parse
import argparse
import numpy as np
import torch
from torch.utils.data import DataLoader
from tqdm import tqdm
from dataset import Badminton_Dataset
from utils import *
parser = argparse.ArgumentParser()
parser.add_argument('--model_file', type=str, default='TrackNetV2/model_best.pt')
parser.add_argument('--batch_size', type=int, default=20)
parser.add_argument('--tolerance', type=float, default=4)
parser.add_argument('--save_dir', type=str, default='output')
parser.add_argument('--debug', action='store_true', default=False)
parser.add_argument('--analyze', action='store_true', default=False)
args = parser.parse_args()
model_file = args.model_file
batch_size = args.batch_size
tolerance = args.tolerance
save_dir = args.save_dir
ckpt_dir = model_file.split('/model')[0]
debug = args.debug
analyze = args.analyze
if not os.path.exists(save_dir):
os.makedirs(save_dir)
checkpoint = torch.load(model_file)
param_dict = checkpoint['param_dict']
model_name = param_dict['model_name']
num_frame = param_dict['num_frame']
input_type = param_dict['input_type']
if __name__ == '__main__':
if not analyze:
# Load dataset
print(f'Data dir: {data_dir}')
print(f'Data input type: {input_type}')
train_dataset = Badminton_Dataset(root_dir=data_dir, split='train', mode=input_type, num_frame=num_frame, slideing_step=1, debug=debug)
test_dataset = Badminton_Dataset(root_dir=data_dir, split='test', mode=input_type, num_frame=num_frame, slideing_step=1, debug=debug)
train_loader = DataLoader(train_dataset, batch_size=batch_size, shuffle=False, drop_last=False)
test_loader = DataLoader(test_dataset, batch_size=batch_size, shuffle=False, drop_last=False)
# Load model
model = get_model(model_name, num_frame, input_type).cuda()
model.load_state_dict(checkpoint['model_state_dict'])
model.eval()
if not os.path.exists(f'{save_dir}/eval_analysis.json'):
print(f'Tolerance Value: {tolerance}')
print('Train evaluation')
accuracy, precision, recall, TP, TN, FP1, FP2, FN = evaluation(model, train_loader, tolerance, input_type)
train_res_dict = {'TP':TP, 'TN': TN, 'FP1': FP1, 'FP2': FP2, 'FN': FN}
print('Test evaluation')
accuracy, precision, recall, TP, TN, FP1, FP2, FN = evaluation(model, test_loader, tolerance, input_type)
test_res_dict = {'TP':TP, 'TN': TN, 'FP1': FP1, 'FP2': FP2, 'FN': FN}
eval_dict = dict(train=train_res_dict, test=test_res_dict, data_dir=data_dir)
with open(f'{save_dir}/eval_analysis.json', 'w') as f:
json.dump(eval_dict, f, indent=2)
if not os.path.exists(f'{save_dir}/eval_res.json'):
eval_dict = json.load(open(f'{save_dir}/eval_analysis.json'))
train_res_dict = eval_dict['train']
TP, TN, FP1, FP2, FN = len(train_res_dict['TP']), len(train_res_dict['TN']), len(train_res_dict['FP1']), len(train_res_dict['FP2']), len(train_res_dict['FN'])
accuracy, precision, recall = get_metric(TP, TN, FP1, FP2, FN)
train_res = {'accuracy': accuracy, 'precision': precision, 'recall': recall, 'TP':TP, 'TN': TN, 'FP1': FP1, 'FP2': FP2, 'FN': FN}
test_res_dict = eval_dict['test']
TP, TN, FP1, FP2, FN = len(test_res_dict['TP']), len(test_res_dict['TN']), len(test_res_dict['FP1']), len(test_res_dict['FP2']), len(test_res_dict['FN'])
accuracy, precision, recall = get_metric(TP, TN, FP1, FP2, FN)
test_res = {'accuracy': accuracy, 'precision': precision, 'recall': recall, 'TP':TP, 'TN': TN, 'FP1': FP1, 'FP2': FP2, 'FN': FN}
res_dict = dict(train=train_res, test=test_res)
with open(f'{save_dir}/eval_res.json', 'w') as f:
json.dump(res_dict, f, indent=2)
loss_list = checkpoint['loss_list']
test_acc_dict = checkpoint['test_acc']
plot_result(loss_list, None, test_acc_dict, num_frame, save_dir, model_name)
else:
# Gather statistic from evaluation results
assert os.path.exists(f'{save_dir}/eval_analysis.json')
train_data = np.load(f'{data_dir}/f{num_frame}_s1_train.npz')
test_data = np.load(f'{data_dir}/f{num_frame}_s1_test.npz')
eval_dict = json.load(open(f'{save_dir}/eval_analysis.json'))
train_res_dict = eval_dict['train']
test_res_dict = eval_dict['test']
train_FN = np.array(train_res_dict['FN'])
train_FP1 = np.array(train_res_dict['FP1'])
train_FP2 = np.array(train_res_dict['FP2'])
train_FN_frames = get_eval_frame_pathes(train_FN, train_data)
train_FP1_frames = get_eval_frame_pathes(train_FP1, train_data)
train_FP2_frames = get_eval_frame_pathes(train_FP2, train_data)
train_FN_res = get_eval_statistic(data_dir, train_FN_frames)
train_FP1_res = get_eval_statistic(data_dir, train_FP1_frames)
train_FP2_res = get_eval_statistic(data_dir, train_FP2_frames)
test_FN = np.array(test_res_dict['FN'])
test_FP1 = np.array(test_res_dict['FP1'])
test_FP2 = np.array(test_res_dict['FP2'])
test_FN_frames = get_eval_frame_pathes(test_FN, test_data)
test_FP1_frames = get_eval_frame_pathes(test_FP1, test_data)
test_FP2_frames = get_eval_frame_pathes(test_FP2, test_data)
test_FN_res = get_eval_statistic(data_dir, test_FN_frames)
test_FP1_res = get_eval_statistic(data_dir, test_FP1_frames)
test_FP2_res = get_eval_statistic(data_dir, test_FP2_frames)
statistic_dict = {
'test_statistic':{'FN':test_FN_res,
'FP1': test_FP1_res,
'FP2': test_FP2_res},
'train_statistic':{'FN':train_FN_res,
'FP1': train_FP1_res,
'FP2': train_FP2_res},
'test_path': {'FN':test_FN_frames,
'FP1': test_FP1_frames,
'FP2': test_FP2_frames},
'train_path': {'FN':train_FN_frames,
'FP1': train_FP1_frames,
'FP2': train_FP2_frames}
}
eval_dict['statistic'] = statistic_dict
with open(f'{save_dir}/eval_analysis.json', 'w') as f:
json.dump(eval_dict, f, indent=2)
plot_eval_statistic(test_FN_res, test_FP1_res, test_FP2_res, 'test', f'{save_dir}/error_analysis_test', figsize=(12, 5))
plot_eval_statistic(train_FN_res, train_FP1_res, train_FP2_res, 'train', f'{save_dir}/error_analysis_train', figsize=(40, 5))