for method, dir in zip(method_list, dir_list):
seq_results = []
for seq in seq_list:
print('sequence: {}'.format(seq))
results_dir = dir + '/results/scale/'
config = load_obj('{}/config'.format(dir))
config['test_seq'] = [seq]
config['data_dir'] = path_to_dset_downsized+config['img_resolution'] + '_res/'
dpc = config['dpc']
mode = config['pose_output_type']
if dpc:
prefix = 'dpc'
else:
prefix=''
test_dset_loaders, _, _ = data_and_model_loader(config, None, None, seq=seq)
data = load_obj('{}/{}_plane_fit'.format(results_dir, config['test_seq'][0]))
dist_to_plane = data['dist_to_plane']
fwd_pose_vec1 = data['fwd_pose_vec1']
fwd_pose_vec2 = data['fwd_pose_vec2']
inv_pose_vec1 = data['inv_pose_vec1']
inv_pose_vec2 = data['inv_pose_vec2']
gt_pose_vec = data['gt_pose_vec']
vo_pose_vec = data['vo_pose_vec']
num_inliers = data['num_inliers']
normals = data['normal']
if config['dpc'] == False:
prefix = ''
if config['dpc'] == True:
## uncomment for KITTI to oxford
# config['estimator'] = 'orbslam'
config['load_stereo'] = False
config['augment_motion'] = False
config['augment_backwards'] = False
config['test_seq'] = [seq]
config['minibatch'] = 1
config['load_pretrained'] = True
dpc = config['dpc']
mode = config['pose_output_type']
if dpc:
prefix = 'dpc'
else:
prefix = ''
test_dset_loaders, models, device = data_and_model_loader(
config, pretrained_depth_path, pretrained_pose_path, seq=seq)
depth_model, pose_model = models[0], models[1]
os.makedirs('scale_results/plane_imgs', exist_ok=True)
os.makedirs('scale_results/plane_imgs/{}'.format(config['test_seq'][0]),
exist_ok=True)
def pf(img_list, depth, intrinsics, img_for_plotting):
img, d, K = img_list, depth, intrinsics
img = img[int(v_crop_min_num * img_list.shape[0] /
v_crop_min_den):int(v_crop_max_num * img_list.shape[0] /
v_crop_max_den),
int(u_crop_min_num * img_list.shape[1] /
u_crop_min_den):int(u_crop_max_num * img_list.shape[1] /
u_crop_max_den)]
d = d[int(v_crop_min_num * img_list.shape[0] /
v_crop_min_den):int(v_crop_max_num * img_list.shape[0] /
splits_dir = '{}/data/splits/{}'.format(path_to_ws, benchmark)
config = load_obj('{}/config'.format(dir))
pretrained_depth_path = glob.glob(
'{}/**depth**best-loss-val_seq-**-test_seq-{}**.pth'.format(dir,
''))[0]
pretrained_pose_path = glob.glob(
'{}/**pose**best-loss-val_seq-**-test_seq-{}**.pth'.format(dir, ''))[0]
config['data_dir'] = path_to_dset_downsized + config[
'img_resolution'] + '_res/' #
config['minibatch'] = 6
config['load_pretrained'] = True
_, models, device = data_and_model_loader(config,
pretrained_depth_path,
pretrained_pose_path,
seq=None)
depth_model, pose_model = models[0], models[1]
if plane_rescaling == True:
### Plane Model
from models.plane_net import PlaneModel, scale_recovery
from losses import Plane_Height_loss
plane_loss = Plane_Height_loss(config)
plane_model = PlaneModel(config).to(config['device'])
pretrained_plane_path = glob.glob(
'../{}/**plane**.pth'.format(pretrained_plane_dir))[0]
plane_model.load_state_dict(torch.load(pretrained_plane_path))
plane_model.train(False).eval()
test_dset = KittiLoaderPytorch(
