exit(-1)
# read the camera parameter of this dataset
with open(osp.join(dataset_path, 'camera_parameter.pickle'),
'rb') as f:
camera_parameter = pickle.load(f)
# using preprocessed 2D poses or using CPN to predict 2D pose
if args.dumped_dir:
test_dataset = PreprocessedDataset(args.dumped_dir[dataset_idx])
logger.info(
f"Using pre-processed datasets {args.dumped_dir[dataset_idx]} for quicker evaluation"
)
else:
test_dataset = BaseDataset(dataset_path, test_range)
test_loader = DataLoader(test_dataset,
batch_size=1,
pin_memory=True,
num_workers=6,
shuffle=False)
pose_in_range = export(test_model,
test_loader,
is_info_dicts=bool(args.dumped_dir),
show=True)
with open(
osp.join(
model_cfg.root_dir, 'result',
time.strftime(
str(model_cfg.testing_on) + "_%Y_%m_%d_%H_%M",
# logging.info ( f'A pose proposal deleted on {img_id}:{person}' )
sub_imageid = list ()
pass
chosen_img.append ( sub_imageid )
return multi_pose3d, chosen_img
if __name__ == '__main__':
import pickle
import scipy.io as scio
from src.models.model_config import model_cfg
from glob import glob
from tqdm import tqdm
from src.m_utils.base_dataset import BaseDataset
from torch.utils.data import DataLoader, Subset
import random
est = MultiEstimator ( model_cfg, debug=False )
with open ( osp.join ( model_cfg.shelf_path, 'camera_parameter.pickle' ), 'rb' ) as f:
test_camera_parameter = pickle.load ( f )
test_dataset = BaseDataset ( model_cfg.shelf_path, range ( 300, 600 ) )
test_dataset = Subset ( test_dataset, random.sample ( range ( 300 ), 50 ) )
test_loader = DataLoader ( test_dataset, batch_size=1, pin_memory=True, num_workers=12, shuffle=False )
for imgs in tqdm ( test_loader ):
this_imgs = list ()
for img_batch in imgs:
this_imgs.append ( img_batch.squeeze ().numpy () )
poses3d = est.predict ( imgs=this_imgs, camera_parameter=test_camera_parameter, show=False,
template_name='Shelf' )
# print ( poses3d )
if __name__ == '__main__':
import pickle
import scipy.io as scio
from src.models.model_config import model_cfg
from glob import glob
from tqdm import tqdm
from src.m_utils.base_dataset import BaseDataset
from torch.utils.data import DataLoader, Subset
import random
import time
est = MultiEstimator(model_cfg, debug=False)
with open(osp.join(model_cfg.shelf_path, 'camera_parameter.pickle'),
'rb') as f:
test_camera_parameter = pickle.load(f)
test_dataset = BaseDataset(model_cfg.lit_path, range(210, 250))
# test_dataset = Subset ( test_dataset, random.sample ( range ( 300 ), 50 ) )
test_loader = DataLoader(test_dataset,
batch_size=1,
pin_memory=True,
num_workers=6,
shuffle=False)
output = []
for imgs in tqdm(test_loader):
this_imgs = list()
for img_batch in imgs:
this_imgs.append(img_batch.squeeze().numpy())
info_dict = est.predict(imgs=this_imgs,
camera_parameter=test_camera_parameter,
show=False,
template_name='Unified')