Train CNN, MLP, ResNet18, RNN, LSTM on CSI data to classify 7 behaviors.walk, stand, sit, run, fall, bow, null
python==3.11.3
torch==2.0.0+cu117
pandas==2.0.2
CSI data of 30*900 every sample,there exist 600 samples.You can change to your own dataset by modifing dataset_process.py
and main.WifiDataset()
.
Dataset used here is not offered!
python main.py --model MLP --epoch 50 --gpu cuda:0 --batch_size 128
For --model
, you can choose from MLP, ResNet18, RNN, LSTM, CNN_GRU
.
If you want to train model from scratch, add --fromScratch
parameter, for example:
python main.py --model MLP --epoch 50 --gpu cuda:0 --batch_size 128 --fromScratch
Or using follow to train a model from checkpoint:
python main.py --model MLP --epoch 50 --gpu cuda:0 --batch_size 128 --checkpoint ./checkpoint/checkpoint.pth
Only inference using saved checkpoint:
python main.py --model MLP --checkpoint ./checkpoint/checkpoint.pth