59 lines
2.2 KiB
Python
59 lines
2.2 KiB
Python
from collections import OrderedDict
|
|
|
|
from ..data.loaders.dl_config import DatabaseStruct
|
|
|
|
MODELS_URL = {'wflw': 'https://huggingface.co/aprados/spiga/resolve/main/spiga_wflw.pt?download=true',
|
|
'300wpublic': 'https://huggingface.co/aprados/spiga/resolve/main/spiga_300wpublic.pt?download=true',
|
|
'300wprivate': 'https://huggingface.co/aprados/spiga/resolve/main/spiga_300wprivate.pt?download=true',
|
|
'merlrav': 'https://huggingface.co/aprados/spiga/resolve/main/spiga_merlrav.pt?download=true',
|
|
'cofw68': 'https://huggingface.co/aprados/spiga/resolve/main/spiga_300wprivate.pt?download=true'}
|
|
|
|
|
|
class ModelConfig(object):
|
|
|
|
def __init__(self, dataset_name=None, load_model_url=True):
|
|
# Model configuration
|
|
self.model_weights = None
|
|
self.model_weights_path = None
|
|
self.load_model_url = load_model_url
|
|
self.model_weights_url = None
|
|
# Pretreatment
|
|
self.focal_ratio = 1.5 # Camera matrix focal length ratio.
|
|
self.target_dist = 1.6 # Target distance zoom in/out around face.
|
|
self.image_size = (256, 256)
|
|
# Outputs
|
|
self.ftmap_size = (64, 64)
|
|
# Dataset
|
|
self.dataset = None
|
|
|
|
if dataset_name is not None:
|
|
self.update_with_dataset(dataset_name)
|
|
|
|
def update_with_dataset(self, dataset_name):
|
|
|
|
config_dict = {'dataset': DatabaseStruct(dataset_name),
|
|
'model_weights': 'spiga_%s.pt' % dataset_name}
|
|
|
|
if dataset_name == 'cofw68': # Test only
|
|
config_dict['model_weights'] = 'spiga_300wprivate.pt'
|
|
|
|
if self.load_model_url:
|
|
config_dict['model_weights_url'] = MODELS_URL[dataset_name]
|
|
|
|
self.update(config_dict)
|
|
|
|
def update(self, params_dict):
|
|
state_dict = self.state_dict()
|
|
for k, v in params_dict.items():
|
|
if k in state_dict or hasattr(self, k):
|
|
setattr(self, k, v)
|
|
else:
|
|
raise Warning('Unknown option: {}: {}'.format(k, v))
|
|
|
|
def state_dict(self):
|
|
state_dict = OrderedDict()
|
|
for k in self.__dict__.keys():
|
|
if not k.startswith('_'):
|
|
state_dict[k] = getattr(self, k)
|
|
return state_dict
|