Torchsense is a library for sensor data processing with PyTorch. It provides I/O, signal and data processing functions, datasets, model implementations and application components.
# clone project
git clone https://github.com/xibrer/torchsense.git
# pip install
pip install torchsenseThe directory structure of your dataset folder looks like this:
data/
├── class_x
│ ├── xxx.ext
│ ├── xxy.ext
│ └── ...
│ └── xxz.ext
└── class_y
├── 123.ext
├── nsdf3.ext
└── ...
└── asd932_.ext
Show details
you can only use our data loader
- the only you need to input are
params([input_key1,...],[target_key])anddata_path
from torchsense.trainer import Trainer
from torchsense.datasets.custom import SensorFolder
from torch.utils.data import DataLoader
from torchsense.models.gan_g import Generator
from torchsense import transforms as T
from torchaudio.transforms import Spectrogram
def train():
# data part
data_path = "data1"
transform1 = T.Compose([
T.ToTensor(),
T.Normalize(-1, 1),
Spectrogram(n_fft=512, hop_length=160, win_length=256, power=1),
])
transform2 = T.Compose([
T.ToTensor(),
T.Interpolate(5000),
Spectrogram(n_fft=100, hop_length=10, win_length=100, power=1),
])
data = SensorFolder(root=data_path,
params=(["acc[2]", "mix_mic"], ["mic"]),
transform=[transform2, transform1],
target_transform=transform1)
train_set, test_set = data.train_test_split(0.5)
train_loader = DataLoader(train_set, batch_size=1, shuffle=True,
drop_last=True, num_workers=0, pin_memory=True)
val_loader = DataLoader(test_set, batch_size=1, shuffle=False,
drop_last=True, num_workers=0)
# model part
model = Generator()
# training part
trainer = Trainer(model, max_epochs=5, task="m")
trainer.fit(train_loader, val_loader)
if __name__ == "__main__":
train()Show details
you can only use our trainer or model
- the only you need to input are
modelordataset
import torch
from torchsense.trainer import Trainer
from torchsense.models.cnn4 import CNN4
def train():
from torchvision.datasets import MNIST
from torchvision.transforms import ToTensor
train_set = MNIST(root="./tmp/data1/MNIST", train=True, transform=ToTensor(), download=True)
val_set = MNIST(root="./tmp/data1/MNIST", train=False, transform=ToTensor(), download=False)
train_loader = torch.utils.data.DataLoader(
train_set, batch_size=64, shuffle=True, pin_memory=torch.cuda.is_available(), num_workers=0
)
val_loader = torch.utils.data.DataLoader(
val_set, batch_size=64, shuffle=False, pin_memory=torch.cuda.is_available(), num_workers=0
)
model = CNN4()
trainer = Trainer(model, max_epochs=5)
trainer.fit(train_loader, val_loader)u-net parts reference milesial