class SampleDataset(Dataset):
def __init__(self, file_pth: str, train=True, random_state=None):
super(Dataset, self).__init__()
self.sp = Spectrogram()
self.file_pth = file_pth
self.train = train
self.random_state = None
self.train = self.read_data()
def __getitem__(self, index):
# datasets[0][index][0]
stage_layer, degc_layer = self.feature_layer(self.train[index][2])
img = self.sp.spec_array(self.train[index][0])
img = np.concatenate((img, stage_layer), axis=0)
img = np.concatenate((img, degc_layer), axis=0)
img = torch.tensor(img, dtype=torch.float32)
target = self.train[index][1]
target = torch.tensor(target)
return img, target
def __len__(self):
return len(self.train)
def feature_layer(self, features):
stage = int(re.findall("\d+", features[0])[0])
degc = float(features[1])
stage_layer = np.full((224, 224), stage, dtype=np.int8)
degc_layer = np.full((224, 224), degc, dtype=np.float)
stage_layer = stage_layer[np.newaxis, :, :]
degc_layer = degc_layer[np.newaxis, :, :]
return stage_layer, degc_layer
def read_data(self):
datas = []
#self.file = "./sample_data.txt"
with open(self.file_pth, "r") as f:
header = f.readline()
while 1:
line = f.readline()
if not line:
break
tmp = line.strip().split('\t')
freq = list(map(float, tmp[4:]))
features = tmp[2:4]
# freq = list(map(float, tmp[1:]))
# print(freq[-1])
label = int(tmp[0])
datas.append([freq, label, features])
return datas
class TestDataset(Dataset):
def __init__(self, file_pth: str, train=True, random_state=None):
super(Dataset, self).__init__()
self.sp = Spectrogram()
self.file_pth = file_pth
self.train = train
self.random_state = None
self.data = self.read_data()
datasets = train_test_split(self.data)
if train:
self.train = datasets[0]
else:
# actually this is test dataset
self.train = datasets[1]
def __getitem__(self, index):
# datasets[0][index][0]
img = self.sp.spec_array(self.train[index][0])
img = torch.tensor(img, dtype=torch.float32)
target = self.train[index][1]
target = torch.tensor(target)
return img, target
def __len__(self):
return len(self.train)
def read_data(self):
datas = []
#self.file = "./sample_data.txt"
with open(self.file_pth, "r") as f:
header = f.readline()
while 1:
line = f.readline()
if not line:
break
tmp = line.strip().split('\t')
# freq = list(map(float, tmp[4:]))
freq = list(map(float, tmp[1:]))
# print(freq[-1])
label = int(tmp[0])
datas.append([freq,label])
return datas