-
Notifications
You must be signed in to change notification settings - Fork 67
/
train.py
105 lines (88 loc) · 3.73 KB
/
train.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
import numpy as np
import math
import random
import os
import torch
from path import Path
from source import model
from source import dataset
from source import utils
from source.args import parse_args
from torchvision import transforms
from torch.utils.data import Dataset, DataLoader
random.seed = 42
def pointnetloss(outputs, labels, m3x3, m64x64, alpha = 0.0001):
criterion = torch.nn.NLLLoss()
bs=outputs.size(0)
id3x3 = torch.eye(3, requires_grad=True).repeat(bs,1,1)
id64x64 = torch.eye(64, requires_grad=True).repeat(bs,1,1)
if outputs.is_cuda:
id3x3=id3x3.cuda()
id64x64=id64x64.cuda()
diff3x3 = id3x3-torch.bmm(m3x3,m3x3.transpose(1,2))
diff64x64 = id64x64-torch.bmm(m64x64,m64x64.transpose(1,2))
return criterion(outputs, labels) + alpha * (torch.norm(diff3x3)+torch.norm(diff64x64)) / float(bs)
def train(args):
path = Path(args.root_dir)
folders = [dir for dir in sorted(os.listdir(path)) if os.path.isdir(path/dir)]
classes = {folder: i for i, folder in enumerate(folders)};
train_transforms = transforms.Compose([
utils.PointSampler(1024),
utils.Normalize(),
utils.RandRotation_z(),
utils.RandomNoise(),
utils.ToTensor()
])
device = torch.device('cuda:0' if torch.cuda.is_available() else 'cpu')
print(device)
pointnet = model.PointNet()
pointnet.to(device)
optimizer = torch.optim.Adam(pointnet.parameters(), lr=args.lr)
train_ds = dataset.PointCloudData(path, transform=train_transforms)
valid_ds = dataset.PointCloudData(path, valid=True, folder='test', transform=train_transforms)
print('Train dataset size: ', len(train_ds))
print('Valid dataset size: ', len(valid_ds))
print('Number of classes: ', len(train_ds.classes))
train_loader = DataLoader(dataset=train_ds, batch_size=args.batch_size, shuffle=True)
valid_loader = DataLoader(dataset=valid_ds, batch_size=args.batch_size*2)
try:
os.mkdir(args.save_model_path)
except OSError as error:
print(error)
print('Start training')
for epoch in range(args.epochs):
pointnet.train()
running_loss = 0.0
for i, data in enumerate(train_loader, 0):
inputs, labels = data['pointcloud'].to(device).float(), data['category'].to(device)
optimizer.zero_grad()
outputs, m3x3, m64x64 = pointnet(inputs.transpose(1,2))
loss = pointnetloss(outputs, labels, m3x3, m64x64)
loss.backward()
optimizer.step()
# print statistics
running_loss += loss.item()
if i % 10 == 9: # print every 10 mini-batches
print('[Epoch: %d, Batch: %4d / %4d], loss: %.3f' %
(epoch + 1, i + 1, len(train_loader), running_loss / 10))
running_loss = 0.0
pointnet.eval()
correct = total = 0
# validation
if valid_loader:
with torch.no_grad():
for data in valid_loader:
inputs, labels = data['pointcloud'].to(device).float(), data['category'].to(device)
outputs, __, __ = pointnet(inputs.transpose(1,2))
_, predicted = torch.max(outputs.data, 1)
total += labels.size(0)
correct += (predicted == labels).sum().item()
val_acc = 100. * correct / total
print('Valid accuracy: %d %%' % val_acc)
# save the model
checkpoint = Path(args.save_model_path)/'save_'+str(epoch)+'.pth'
torch.save(pointnet.state_dict(), checkpoint)
print('Model saved to ', checkpoint)
if __name__ == '__main__':
args = parse_args()
train(args)