class ModelInterface:
def __init__(self):
self.features = defaultdict(list)
self.gmmset = GMMSet()
def enroll(self, name, fs, signal):
feat = get_feature(fs, signal)
self.features[name].extend(feat)
def train_full(self):
self.gmmset = GMMSet()
start_time = time.time()
print('==============', len(self.features.items()))
for name, feats in self.features.items():
try:
self.gmmset.fit_new(feats, name)
except Exception as e:
print("%s failed" % (name))
print(time.time() - start_time, " seconds")
def train_single(self, lable):
start_time = time.time()
try:
self.gmmset.fit_new(self.features.__getitem__(lable), lable)
except Exception as e:
print("%s failed" % (lable))
print(time.time() - start_time, " seconds")
def dump(self, fname):
""" dump all models to file"""
self.gmmset.before_pickle()
with open(fname, 'wb') as f:
pickle.dump(self, f, -1)
self.gmmset.after_pickle()
def predict(self, fs, signal):
"""
return a label (name)
"""
try:
feat = get_feature(fs, signal)
except Exception as e:
print(e)
return self.gmmset.predict_one(feat)
def verify(self, fs, signal, personid):
"""
return a label (name)
"""
try:
feat = get_feature(fs, signal)
except Exception as e:
print(e)
return self.gmmset.verify(feat, personid)
@staticmethod
def load(fname):
""" load from a dumped model file"""
with open(fname, 'rb') as f:
R = pickle.load(f)
R.gmmset.after_pickle()
return R
