def get_generator(model, cfg, device, **kwargs):
''' Returns the generator object.
Args:
model (nn.Module): Occupancy Network model
cfg (dict): imported yaml config
device (device): pytorch device
'''
preprocessor = config.get_preprocessor(cfg, device=device)
generator = generation.Generator3D(
model,
device=device,
threshold=cfg['test']['threshold'],
resolution0=cfg['generation']['resolution_0'],
upsampling_steps=cfg['generation']['upsampling_steps'],
sample=cfg['generation']['use_sampling'],
refinement_step=cfg['generation']['refinement_step'],
simplify_nfaces=cfg['generation']['simplify_nfaces'],
preprocessor=preprocessor,
pnet_point_scale=cfg['trainer']['pnet_point_scale'],
is_explicit_mesh=cfg['generation'].get('is_explicit_mesh', False),
is_skip_surface_mask_generation_time=cfg['generation'].get(
'is_skip_surface_mask_generation_time', False),
is_just_measuring_time=cfg['generation'].get('is_just_measuring_time',
False),
is_fit_to_gt_loc_scale=cfg['generation'].get('is_fit_to_gt_loc_scale',
False),
**cfg['generation'].get('mesh_kwargs', {}),
)
return generator
def get_generator(model, cfg):
''' Returns the generator object.
Args:
model (tf.keras.Model): Occupancy Network model
cfg (dict): imported yaml config
'''
preprocessor = config.get_preprocessor(cfg)
generator = generation.Generator3D(
model,
threshold=cfg['test']['threshold'],
resolution0=cfg['generation']['resolution_0'],
upsampling_steps=cfg['generation']['upsampling_steps'],
sample=cfg['generation']['use_sampling'],
refinement_step=cfg['generation']['refinement_step'],
simplify_nfaces=cfg['generation']['simplify_nfaces'],
preprocessor=preprocessor,
)
return generator
def get_generator(model, cfg, device, **kwargs):
''' Returns the generator object.
Args:
model (nn.Module): Occupancy Network model
cfg (dict): imported yaml config
device (device): pytorch device
'''
preprocessor = config.get_preprocessor(cfg, device=device)
input_type = cfg['data']['input_type']
generator_params = {
'device': device,
'threshold': cfg['test']['threshold'],
'resolution0': cfg['generation']['resolution_0'],
'upsampling_steps': cfg['generation']['upsampling_steps'],
'sample': cfg['generation']['use_sampling'],
'refinement_step' :cfg['generation']['refinement_step'],
'simplify_nfaces' :cfg['generation']['simplify_nfaces'],
'preprocessor' :preprocessor,
'input_type' :input_type,
}
if input_type == 'depth_pred':
generator_params['use_gt_depth_map'] = cfg['training']['use_gt_depth']
if 'pred_with_img' in cfg['model']:
generator_params['with_img'] = cfg['model']['pred_with_img']
if 'depth_pointcloud_transfer' in cfg['model']:
generator_params['depth_pointcloud_transfer'] = cfg['model']['depth_pointcloud_transfer']
if 'use_local_feature' in cfg['model']:
generator_params['local'] = cfg['model']['use_local_feature']
generator = generation.Generator3D(
model,
**generator_params
)
return generator
def get_generator(model, cfg, device, **kwargs):
''' Returns the generator object.
Args:
model (nn.Module): Occupancy Network model
cfg (dict): imported yaml config
device (device): pytorch device
'''
preprocessor = config.get_preprocessor(cfg, device=device)
generator = generation.Generator3D(
model,
device=device,
threshold=cfg['test']['threshold'],
resolution0=cfg['generation']['resolution_0'],
upsampling_steps=cfg['generation']['upsampling_steps'],
sample=cfg['generation']['use_sampling'],
refinement_step=cfg['generation']['refinement_step'],
simplify_nfaces=cfg['generation']['simplify_nfaces'],
preprocessor=preprocessor,
)
return generator
