-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathget_metrics.py
142 lines (109 loc) · 4.92 KB
/
get_metrics.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
#!/usr/bin/env python
"""
Calculates the metrics for testing ETH/UCY outputs.
Input files with input and output trajectories
"""
__author__ = "Aamir Hasan"
__version__ = "1.0"
__email__ = "hasanaamir215@gmail.com; aamirh2@illinois.edu"
from argparse import ArgumentParser
from os.path import exists, join
from os import makedirs, listdir
import pickle
import numpy as np
import csv
def get_length(trajectory):
end_pos = 20
for j in range(20):
if trajectory[j, 0] == -2 and trajectory[j, 1] == -2:
end_pos = j
return end_pos
def calculate_ADE(input_trajectory, output_trajectory):
diff = output_trajectory - input_trajectory
return np.sqrt((diff**2).sum(axis=-1)).mean()
def calculate_FDE(input_trajectory, output_trajectory):
diff = output_trajectory[-1, :] - input_trajectory[-1, :]
return np.sqrt((diff**2).sum(axis=-1))
def process_file(filename):
# each file contains
# trajectories, output_trajectories, valid_peds, N, ped_ids, min_val, max_val, scene_id, dataset_name
# no need to look through dataset name since, all test files come in from a single dataset
num_peds_per_tstep = np.zeros(12)
other_ades = np.zeros(12)
ade = 0
fde = 0
ade_obs = 0
data = pickle.load(open(filename, 'rb'))
# trajectories, output_trajectories, valid_peds, N, ped_ids, min_val, max_val, scene_id, dataset_name
num_peds = data[2]
input_trajectories = data[0].cpu().numpy()
output_trajectories = data[1].cpu().numpy()
max_vals = [1.0208, 1.0278]
min_vals = [-0.99653, -1.0167]
input_trajectories[:, :, 0] = (max_vals[0] - min_vals[0]) * (input_trajectories[:, :, 0] + 1) / 2 + min_vals[0]
input_trajectories[:, :, 1] = (max_vals[1] - min_vals[1]) * (input_trajectories[:, :, 1] + 1) / 2 + min_vals[1]
output_trajectories[:, :, 0] = (max_vals[0] - min_vals[0]) * (output_trajectories[:, :, 0] + 1) / 2 + min_vals[0]
output_trajectories[:, :, 1] = (max_vals[1] - min_vals[1]) * (output_trajectories[:, :, 1] + 1) / 2 + min_vals[1]
# go through all trajectories and calculate
# ADE and FDE - functions
for i in range(num_peds):
end_pos = get_length(input_trajectories[i, :, :])
ade += calculate_ADE(input_trajectories[i, :end_pos, :], output_trajectories[i, :end_pos, :])
fde += calculate_FDE(input_trajectories[i, :end_pos, :], output_trajectories[i, :end_pos, :])
ade_obs += calculate_ADE(input_trajectories[i, 8:end_pos, :], output_trajectories[i, 8:end_pos, :])
for j in range(12):
if 9 + j <= end_pos:
other_ades[j] += calculate_ADE(input_trajectories[i, 8:9+j, :], output_trajectories[i, 8:9+j, :])
num_peds_per_tstep[j] += 1
#ade = ade
#fde = fde / num_peds
return ade, fde, ade_obs, num_peds, data[3], other_ades, num_peds_per_tstep
def parse_args():
"""
Parses the arguments to the executable,
:return: args
:rtype: dictionary of all the arguments passed in
"""
parser = ArgumentParser()
parser.add_argument('--load_path', type=str, default='./save',
help='Path to the directory with the saved trajectories')
parser.add_argument('--csv_save_path', type=str, default='./results',
help='Path to the save the results csv')
parser.add_argument('--save_name', type=str, default='eth_ucy_0',
help='Name of the saved csv')
args = parser.parse_args()
return args
def main():
args = parse_args()
if not exists(args.load_path):
raise ValueError("Saved files do not exist")
if not exists(args.csv_save_path):
makedirs(args.csv_save_path)
csv_out_file = open(join(args.csv_save_path, args.save_name + ".csv"), 'w')
csv_writer = csv.writer(csv_out_file, delimiter=',')
csv_writer.writerow(['filename', 'ade', 'fde', 'ade_obs', 'num_peds', 'total_peds', 'ade_0', 'ade_1', 'ade_2', 'ade_3', 'ade_4', 'ade_5', 'ade_6', 'ade_7', 'ade_8', 'ade_9', 'ade_10', 'ade_11', 'num_0', 'num_1', 'num_2', 'num_3', 'num_4', 'num_5', 'num_6', 'num_7', 'num_8', 'num_9', 'num_10', 'num_11'])
running_ade = 0
running_fde = 0
running_ade_obs = 0
num_files = 0
# go through all files in directory
for filename in listdir(args.load_path):
ade, fde, ade_obs, num_peds, N, all_ades, num_peds_t = process_file(join(args.load_path, filename))
# save to csv
csv_writer.writerow(np.concatenate([[filename, ade, fde, ade_obs, num_peds, N], all_ades, num_peds_t]))
running_ade += ade
running_fde += fde
running_ade_obs += ade_obs
num_files += 1
# Average out for all files in the directory
running_fde /= num_files
running_ade /= num_files
running_ade_obs /= num_files
csv_out_file.close()
# Report
print(f"FDE: {running_fde}")
print(f"ADE: {running_ade}")
print(f"ADE_obs: {running_ade_obs}")
print("Done!")
if __name__ == "__main__":
main()