Repository navigation
Expand file tree
/
Copy pathevaluate.py
More file actions
106 lines (80 loc) · 3.54 KB
/
Copy pathevaluate.py
File metadata and controls
106 lines (80 loc) · 3.54 KB
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
import os
import numpy as np
import torch
from argparse import ArgumentParser
import tqdm
from config import paths, joint_set, fps
import articulate as art
from data import PoseDataset
from model import MobilePoser
from utils import load_yaml
class PoseEvaluator:
def __init__(self, device='cpu'):
self._eval_fn = art.FullMotionEvaluator(paths.smpl_file, joint_mask=torch.tensor([2, 5, 16, 20], device=device), fps=fps, device=device)
def eval(self, pose_p, pose_t, tran_p=None, tran_t=None):
pose_p = pose_p.clone().view(-1, 24, 3, 3)
pose_t = pose_t.clone().view(-1, 24, 3, 3)
tran_p = tran_p.clone().view(-1, 3)
tran_t = tran_t.clone().view(-1, 3)
pose_p[:, joint_set.ignored] = torch.eye(3, device=pose_p.device)
pose_t[:, joint_set.ignored] = torch.eye(3, device=pose_t.device)
errs = self._eval_fn(pose_p, pose_t, tran_p=tran_p, tran_t=tran_t)
return torch.stack([errs[9], errs[3], errs[9], errs[0]*100, errs[7]*100, errs[1]*100, errs[4] / 100, errs[6]])
@staticmethod
def print(errors):
for i, name in enumerate([
'SIP Error (deg)', 'Angular Error (deg)', 'Masked Angular Error (deg)',
'Positional Error (cm)', 'Masked Positional Error (cm)', 'Mesh Error (cm)',
'Jitter Error (100m/s^3)', 'Distance Error (cm)'
]):
print('%s: %.2f (+/- %.2f)' % (name, errors[i, 0], errors[i, 1]))
@torch.no_grad()
def evaluate_pose(model, dataset, num_future_frame=5):
# specify device
device = cfg['device']
# load data
xs, ys = zip(*[
(imu.to(device), (pose_6d.to(device), tran.to(device)))
for imu, pose_6d, joint, tran, vel, contact in dataset
])
# setup Pose Evaluator
evaluator = PoseEvaluator(device=device)
# track errors
offline_errs, online_errs = [], []
model.eval()
with torch.no_grad():
for x, y in tqdm.tqdm(list(zip(xs, ys))):
model.reset()
# offline
pose_p_offline, joint_p_offline, tran_p_offline, _ = model.forward_offline(x.unsqueeze(0))
pose_t_6d, tran_t = y
pose_t = art.math.r6d_to_rotation_matrix(pose_t_6d)
# online
if os.getenv("ONLINE"):
online_results = [model.forward_online(f) for f in torch.cat((x, x[-1].repeat(num_future_frame, 1)))]
pose_p_online, joint_p_online, tran_p_online, _ = [torch.stack(_)[num_future_frame:] for _ in zip(*online_results)]
# store errors
offline_errs.append(evaluator.eval(pose_p_offline, pose_t, tran_p=tran_p_offline, tran_t=tran_t))
if os.getenv("ONLINE"):
online_errs.append(
evaluator.eval(pose_p_online, pose_t, tran_p=tran_p_online, tran_t=tran_t)
)
# print joint errors
print('============== offline ================')
evaluator.print(torch.stack(offline_errs).mean(dim=0))
if os.getenv("ONLINE"):
print('============== online ================')
evaluator.print(torch.stack(online_errs).mean(dim=0))
if __name__ == '__main__':
parser = ArgumentParser()
parser.add_argument('--weights', type=str, required=True)
parser.add_argument('--dataset', type=str, default='test')
args = parser.parse_args()
# load cfg
cfg = load_yaml('evaluate.yaml')
# load model
model = MobilePoser.from_pretrained(cfg, args.weights).to(cfg['device'])
# load dataset
dataset = PoseDataset(cfg, fold='test', evaluate=True)
# evaluate
evaluate_pose(model, dataset)