-
Notifications
You must be signed in to change notification settings - Fork 1
/
eval.py
68 lines (51 loc) · 1.86 KB
/
eval.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
'''
@FileName : eval.py
@EditTime : 2024-04-03 14:41:12
@Author : Buzhen Huang
@Email : buzhenhuang@outlook.com
@Description :
'''
import torch
from cmd_parser import parse_config
from utils.module_utils import seed_worker, set_seed
from modules import init, LossLoader, ModelLoader, DatasetLoader
from utils.eval_utils import HumanEval
# ###########Load config file in debug mode#########
# import sys
# sys.argv = ['','--config=cfg_files/eval.yaml']
def main(**args):
seed = 7
g = set_seed(seed)
# Global setting
is_seq = False
dtype = torch.float32
batchsize = args.get('batchsize')
num_epoch = args.get('epoch')
workers = args.get('worker')
device = torch.device(index=args.get('gpu_index'), type='cuda')
mode = args.get('mode')
# Initialize project setting, e.g., create output folder, load SMPL model
out_dir, logger, smpl = init(dtype=dtype, **args)
# Load loss function
loss = LossLoader(smpl, device=device, **args)
# Load model
model = ModelLoader(dtype=dtype, device=device, out_dir=out_dir, **args)
# create data loader
dataset = DatasetLoader(dtype=dtype, smpl=smpl, **args)
eval_dataset = dataset.load_evalset()
# Load handle function with the task name
task = args.get('task')
exec('from process import %s_eval' %task)
for i, (name, dataset) in enumerate(zip(dataset.testset, eval_dataset)):
eval_loader = torch.utils.data.DataLoader(
dataset,
batch_size=args.get('batchsize'), shuffle=False,
num_workers=args.get('worker'), pin_memory=True, drop_last=False
)
pred, gt = eval('%s_eval' %task)(model, eval_loader, loss, device=device)
evaluator = HumanEval(name)
evaluator(pred, gt)
evaluator.report()
if __name__ == "__main__":
args = parse_config()
main(**args)