def load_data(self):
if self.partition_modus == 'train':
#idx_files = [ids for ids in range(0,12)]
if self.config["proportions"] == 0.2:
path = '/data/sawasthi/NTU/trainData_pose_1/'
dataset = CustomDataSet(path)
elif self.config["proportions"] == 0.5:
path = '/data/sawasthi/NTU/trainData_pose_1/'
dataset = CustomDataSet(path)
elif self.config["proportions"] == 0.75:
path = '/data/sawasthi/NTU/trainData_pose_1/'
dataset = CustomDataSet(path)
elif self.config["proportions"] == 1.0:
path = '/data/sawasthi/NTU/trainData_pose_1/'
#path = 'S:/Datasets/nturgbd_skeletons_s001_to_s017/train/'
#path = 'S:/MS A&R/4th Sem/Thesis/PAMAP2_Dataset/pkl files'
#path = "S:/MS A&R/4th Sem/Thesis/LaRa/OMoCap data/Train_data/"
dataset = CustomDataSet(path)
elif self.partition_modus == 'val':
path = '/data/sawasthi/NTU/trainData_pose_1/'
dataset = CustomDataSet(path)
elif self.partition_modus == 'test':
path = '/data/sawasthi/NTU/trainData_pose_1/'
dataset = CustomDataSet(path)
else:
raise ("Wrong Dataset partition settup")
logging.info(' Dataloader: Processing dataset files ...')
return dataset
def getTrainData(path, batch_size):
train_data = []
#path = '/data/sawasthi/data/trainData/'
#path = 'S:/MS A&R/4th Sem/Thesis/LaRa/IMU data/IMU data/Windows2/'
#while folder_counter < 10:
#some code to get path_to_imgs which is the location of the image folder
train_dataset = CustomDataSet(path)
#all_datasets.append(train_dataset)
#final_dataset = torch.utils.data.ConcatDataset(train_dataset)
train_loader = DataLoader(train_dataset,
shuffle=False,
batch_size=batch_size,
num_workers=0,
pin_memory=True,
drop_last=True)
for idx, input_seq in enumerate(train_loader):
train_data.append(input_seq)
return train_data
optimizer = optim.RMSprop(model.parameters(),
lr=learning_rate,
alpha=0.9,
weight_decay=0.0005,
momentum=0.9)
#lmbda = lambda epoch: 0.95
#scheduler = lr_scheduler.StepLR(optimizer, step_size=1,gamma=0.95)
#scheduler = lr_scheduler.ReduceLROnPlateau(optimizer, patience=5)
#optimizer = optim.SGD(model.parameters(), lr=0.0001, momentum=0.9)
model_path = '/data/sawasthi/LaraMM/model/Laramm_model_cnn.pth'
#model_path = 'S:/MS A&R/4th Sem/Thesis/LaRa/OMoCap data/model.pth'
path = '/data/sawasthi/LaraMM/sequences_train'
#path = 'S:/MS A&R/4th Sem/Thesis/LaRa/IMU data/IMU data/Windows2/'
#path = "S:/MS A&R/4th Sem/Thesis/LaRa/OMoCap data/Train_data/"
train_dataset = CustomDataSet(path)
dataLoader_train = DataLoader(train_dataset,
shuffle=True,
batch_size=batch_size,
num_workers=0,
pin_memory=True,
drop_last=True)
# Validation data
path = '/data/sawasthi/LaraMM/sequences_val'
#path = 'S:/MS A&R/4th Sem/Thesis/LaRa/IMU data/IMU data/Windows/'
#path = "S:/MS A&R/4th Sem/Thesis/LaRa/OMoCap data/Test_data/"
validation_dataset = CustomDataSet(path)
dataLoader_validation = DataLoader(validation_dataset,
shuffle=False,
batch_size=batch_size,
noise = np.random.normal(0, 1, (batch_size, 1, ws, features))
#noise = np.random.normal(0,1,(batch_size,features,ws))
noise = torch.tensor(noise)
noise = noise.float()
#criterion = nn.NLLLoss()
criterion = nn.CrossEntropyLoss()
optimizer = optim.Adam(model.parameters(), lr=0.0001)
#model_path = '/data/sawasthi/data/model/model.pth'
model_path = '/data/sawasthi/data/MoCAP_data/model/model.pth'
#model_path = 'S:/MS A&R/4th Sem/Thesis/LaRa/OMoCap data/'
#path = 'S:/MS A&R/4th Sem/Thesis/LaRa/IMU data/IMU data/Windows2/'
path = '/data/sawasthi/data/trainData/'
#path = '/data/sawasthi/data/MoCAP_data/trainData/'
# path = 'S:/MS A&R/4th Sem/Thesis/LaRa/IMU data/IMU data/Windows2/'
#path = "S:/MS A&R/4th Sem/Thesis/LaRa/OMoCap data/Train_data/"
train_dataset = CustomDataSet(path)
dataLoader_train = DataLoader(train_dataset,
shuffle=True,
batch_size=batch_size,
num_workers=0,
pin_memory=True,
drop_last=True)
print("preparing data for normalisation")
# Normalise the data
value = []
for k in range(999):
temp_list = []
max = -9999
min = 9999
temp_list.append(max)