In [ ]:
%matplotlib inline
%reload_ext autoreload
%autoreload 2
In [ ]:
from fastai.conv_learner import *
from fastai.dataset import *
from pathlib import Path
import json
torch.cuda.set_device(1)
Data¶
Setup¶
In [ ]:
PATH = Path('data/carvana')
list(PATH.iterdir())
Out[ ]:
[PosixPath('data/carvana/train_masks.csv'),
PosixPath('data/carvana/train_masks-128'),
PosixPath('data/carvana/sample_submission.csv'),
PosixPath('data/carvana/train_masks_png'),
PosixPath('data/carvana/train.csv'),
PosixPath('data/carvana/train-128'),
PosixPath('data/carvana/train'),
PosixPath('data/carvana/metadata.csv'),
PosixPath('data/carvana/tmp'),
PosixPath('data/carvana/models'),
PosixPath('data/carvana/train_masks')]
In [ ]:
MASKS_FN = 'train_masks.csv'
META_FN = 'metadata.csv'
TRAIN_DN = 'train'
MASKS_DN = 'train_masks'
In [ ]:
masks_csv = pd.read_csv(PATH/MASKS_FN)
masks_csv.head()
Out[ ]:
| img | rle_mask | |
|---|---|---|
| 0 | 00087a6bd4dc_01.jpg | 879386 40 881253 141 883140 205 885009 17 8850... |
| 1 | 00087a6bd4dc_02.jpg | 873779 4 875695 7 877612 9 879528 12 881267 15... |
| 2 | 00087a6bd4dc_03.jpg | 864300 9 866217 13 868134 15 870051 16 871969 ... |
| 3 | 00087a6bd4dc_04.jpg | 879735 20 881650 26 883315 92 883564 30 885208... |
| 4 | 00087a6bd4dc_05.jpg | 883365 74 883638 28 885262 119 885550 34 88716... |
In [ ]:
meta_csv = pd.read_csv(PATH/META_FN)
meta_csv.head()
Out[ ]:
| id | year | make | model | trim1 | trim2 | |
|---|---|---|---|---|---|---|
| 0 | 0004d4463b50 | 2014.0 | Acura | TL | TL | w/SE |
| 1 | 00087a6bd4dc | 2014.0 | Acura | RLX | RLX | w/Tech |
| 2 | 000aa097d423 | 2012.0 | Mazda | MAZDA6 | MAZDA6 | i Sport |
| 3 | 000f19f6e7d4 | 2016.0 | Chevrolet | Camaro | Camaro | SS |
| 4 | 00144e887ae9 | 2015.0 | Acura | TLX | TLX | SH-AWD V6 w/Advance Pkg |
In [ ]:
def show_img(im, figsize=None, ax=None, alpha=None):
if not ax: fig,ax = plt.subplots(figsize=figsize)
ax.imshow(im, alpha=alpha)
ax.set_axis_off()
return ax
In [ ]:
CAR_ID = '00087a6bd4dc'
In [ ]:
list((PATH/TRAIN_DN).iterdir())[:5]
Out[ ]:
[PosixPath('data/carvana/train/5ab34f0e3ea5_15.jpg'),
PosixPath('data/carvana/train/de3ca5ec1e59_07.jpg'),
PosixPath('data/carvana/train/28d9a149cb02_13.jpg'),
PosixPath('data/carvana/train/36a3f7f77e85_12.jpg'),
PosixPath('data/carvana/train/843763f47895_08.jpg')]
In [ ]:
Image.open(PATH/TRAIN_DN/f'{CAR_ID}_01.jpg').resize((300,200))
Out[ ]:
In [ ]:
list((PATH/MASKS_DN).iterdir())[:5]
Out[ ]:
[PosixPath('data/carvana/train_masks/6c0cd487abcd_03_mask.gif'),
PosixPath('data/carvana/train_masks/351c583eabd6_01_mask.gif'),
PosixPath('data/carvana/train_masks/90fdd8932877_02_mask.gif'),
PosixPath('data/carvana/train_masks/28d9a149cb02_10_mask.gif'),
PosixPath('data/carvana/train_masks/88bc32b9e1d9_14_mask.gif')]
In [ ]:
Image.open(PATH/MASKS_DN/f'{CAR_ID}_01_mask.gif').resize((300,200))
Out[ ]:
In [ ]:
ims = [open_image(PATH/TRAIN_DN/f'{CAR_ID}_{i+1:02d}.jpg') for i in range(16)]
In [ ]:
fig, axes = plt.subplots(4, 4, figsize=(9, 6))
for i,ax in enumerate(axes.flat): show_img(ims[i], ax=ax)
plt.tight_layout(pad=0.1)
Resize and convert¶
In [ ]:
(PATH/'train_masks_png').mkdir(exist_ok=True)
In [ ]:
def convert_img(fn):
fn = fn.name
Image.open(PATH/'train_masks'/fn).save(PATH/'train_masks_png'/f'{fn[:-4]}.png')
In [ ]:
files = list((PATH/'train_masks').iterdir())
with ThreadPoolExecutor(8) as e: e.map(convert_img, files)
Wall time: 27.4 s
In [ ]:
(PATH/'train_masks-128').mkdir(exist_ok=True)
In [ ]:
def resize_mask(fn):
Image.open(fn).resize((128,128)).save((fn.parent.parent)/'train_masks-128'/fn.name)
files = list((PATH/'train_masks_png').iterdir())
with ThreadPoolExecutor(8) as e: e.map(resize_mask, files)
In [ ]:
(PATH/'train-128').mkdir(exist_ok=True)
In [ ]:
def resize_img(fn):
Image.open(fn).resize((128,128)).save((fn.parent.parent)/'train-128'/fn.name)
files = list((PATH/'train').iterdir())
with ThreadPoolExecutor(8) as e: e.map(resize_img, files)
Dataset¶
In [ ]:
TRAIN_DN = 'train-128'
MASKS_DN = 'train_masks-128'
sz = 128
bs = 64
In [ ]:
ims = [open_image(PATH/TRAIN_DN/f'{CAR_ID}_{i+1:02d}.jpg') for i in range(16)]
im_masks = [open_image(PATH/MASKS_DN/f'{CAR_ID}_{i+1:02d}_mask.png') for i in range(16)]
In [ ]:
fig, axes = plt.subplots(4, 4, figsize=(9, 6))
for i,ax in enumerate(axes.flat):
ax = show_img(ims[i], ax=ax)
show_img(im_masks[i][...,0], ax=ax, alpha=0.5)
plt.tight_layout(pad=0.1)
In [ ]:
class MatchedFilesDataset(FilesDataset):
def __init__(self, fnames, y, transform, path):
self.y=y
assert(len(fnames)==len(y))
super().__init__(fnames, transform, path)
def get_y(self, i): return open_image(os.path.join(self.path, self.y[i]))
def get_c(self): return 0
In [ ]:
x_names = np.array([Path(TRAIN_DN)/o for o in masks_csv['img']])
y_names = np.array([Path(MASKS_DN)/f'{o[:-4]}_mask.png' for o in masks_csv['img']])
In [ ]:
len(x_names)//16//5*16
Out[ ]:
1008
In [ ]:
val_idxs = list(range(1008))
((val_x,trn_x),(val_y,trn_y)) = split_by_idx(val_idxs, x_names, y_names)
len(val_x),len(trn_x)
Out[ ]:
(1008, 4080)
In [ ]:
aug_tfms = [RandomRotate(4, tfm_y=TfmType.CLASS),
RandomFlip(tfm_y=TfmType.CLASS),
RandomLighting(0.05, 0.05)]
# aug_tfms = []
In [ ]:
tfms = tfms_from_model(resnet34, sz, crop_type=CropType.NO, tfm_y=TfmType.CLASS, aug_tfms=aug_tfms)
datasets = ImageData.get_ds(MatchedFilesDataset, (trn_x,trn_y), (val_x,val_y), tfms, path=PATH)
md = ImageData(PATH, datasets, bs, num_workers=8, classes=None)
In [ ]:
denorm = md.trn_ds.denorm
x,y = next(iter(md.aug_dl))
x = denorm(x)
In [ ]:
fig, axes = plt.subplots(5, 6, figsize=(12, 10))
for i,ax in enumerate(axes.flat):
ax=show_img(x[i], ax=ax)
show_img(y[i], ax=ax, alpha=0.5)
plt.tight_layout(pad=0.1)
Model¶
In [ ]:
class Empty(nn.Module):
def forward(self,x): return x
models = ConvnetBuilder(resnet34, 0, 0, 0, custom_head=Empty())
learn = ConvLearner(md, models)
learn.summary()
In [ ]:
class StdUpsample(nn.Module):
def __init__(self, nin, nout):
super().__init__()
self.conv = nn.ConvTranspose2d(nin, nout, 2, stride=2)
self.bn = nn.BatchNorm2d(nout)
def forward(self, x): return self.bn(F.relu(self.conv(x)))
In [ ]:
flatten_channel = Lambda(lambda x: x[:,0])
In [ ]:
simple_up = nn.Sequential(
nn.ReLU(),
StdUpsample(512,256),
StdUpsample(256,256),
StdUpsample(256,256),
StdUpsample(256,256),
nn.ConvTranspose2d(256, 1, 2, stride=2),
flatten_channel
)
In [ ]:
models = ConvnetBuilder(resnet34, 0, 0, 0, custom_head=simple_up)
learn = ConvLearner(md, models)
learn.opt_fn=optim.Adam
learn.crit=nn.BCEWithLogitsLoss()
learn.metrics=[accuracy_thresh(0.5)]
In [ ]:
learn.lr_find()
learn.sched.plot()
A Jupyter Widget
94%|█████████▍| 30/32 [00:05<00:00, 5.48it/s, loss=10.6]
In [ ]:
lr=4e-2
In [ ]:
learn.fit(lr,1,cycle_len=5,use_clr=(20,5))
A Jupyter Widget
epoch trn_loss val_loss <lambda>
0 0.124078 0.133566 0.945951
1 0.111241 0.112318 0.954912
2 0.099743 0.09817 0.957507
3 0.090651 0.092375 0.958117
4 0.084031 0.086026 0.963243
Out[ ]:
[0.086025625, 0.96324310824275017]
In [ ]:
learn.save('tmp')
In [ ]:
learn.load('tmp')
In [ ]:
py,ay = learn.predict_with_targs()
In [ ]:
ay.shape
Out[ ]:
(1008, 128, 128)
In [ ]:
show_img(ay[0]);
In [ ]:
show_img(py[0]>0);
In [ ]:
learn.unfreeze()
In [ ]:
learn.bn_freeze(True)
In [ ]:
lrs = np.array([lr/100,lr/10,lr])/4
In [ ]:
learn.fit(lrs,1,cycle_len=20,use_clr=(20,10))
A Jupyter Widget
epoch trn_loss val_loss <lambda>
0 0.06577 0.053292 0.972977
1 0.049475 0.043025 0.982559
2 0.039146 0.035927 0.98337
3 0.03405 0.031903 0.986982
4 0.029788 0.029065 0.987944
5 0.027374 0.027752 0.988029
6 0.026041 0.026718 0.988226
7 0.024302 0.025927 0.989512
8 0.022921 0.026102 0.988276
9 0.021944 0.024714 0.989537
10 0.021135 0.0241 0.990628
11 0.020494 0.023367 0.990652
12 0.01988 0.022961 0.990989
13 0.019241 0.022498 0.991014
14 0.018697 0.022492 0.990571
15 0.01812 0.021771 0.99105
16 0.017597 0.02183 0.991365
17 0.017192 0.021434 0.991364
18 0.016768 0.021383 0.991643
19 0.016418 0.021114 0.99173
Out[ ]:
[0.021113895, 0.99172959849238396]
In [ ]:
learn.save('0')
In [ ]:
x,y = next(iter(md.val_dl))
py = to_np(learn.model(V(x)))
In [ ]:
ax = show_img(denorm(x)[0])
show_img(py[0]>0, ax=ax, alpha=0.5);
In [ ]:
ax = show_img(denorm(x)[0])
show_img(y[0], ax=ax, alpha=0.5);
512x512¶
In [ ]:
TRAIN_DN = 'train'
MASKS_DN = 'train_masks_png'
sz = 512
bs = 16
In [ ]:
x_names = np.array([Path(TRAIN_DN)/o for o in masks_csv['img']])
y_names = np.array([Path(MASKS_DN)/f'{o[:-4]}_mask.png' for o in masks_csv['img']])
In [ ]:
((val_x,trn_x),(val_y,trn_y)) = split_by_idx(val_idxs, x_names, y_names)
len(val_x),len(trn_x)
Out[ ]:
(1008, 4080)
In [ ]:
tfms = tfms_from_model(resnet34, sz, crop_type=CropType.NO, tfm_y=TfmType.CLASS, aug_tfms=aug_tfms)
datasets = ImageData.get_ds(MatchedFilesDataset, (trn_x,trn_y), (val_x,val_y), tfms, path=PATH)
md = ImageData(PATH, datasets, bs, num_workers=8, classes=None)
In [ ]:
denorm = md.trn_ds.denorm
x,y = next(iter(md.aug_dl))
x = denorm(x)
In [ ]:
fig, axes = plt.subplots(4, 4, figsize=(10, 10))
for i,ax in enumerate(axes.flat):
ax=show_img(x[i], ax=ax)
show_img(y[i], ax=ax, alpha=0.5)
plt.tight_layout(pad=0.1)
In [ ]:
simple_up = nn.Sequential(
nn.ReLU(),
StdUpsample(512,256),
StdUpsample(256,256),
StdUpsample(256,256),
StdUpsample(256,256),
nn.ConvTranspose2d(256, 1, 2, stride=2),
flatten_channel
)
In [ ]:
models = ConvnetBuilder(resnet34, 0, 0, 0, custom_head=simple_up)
learn = ConvLearner(md, models)
learn.opt_fn=optim.Adam
learn.crit=nn.BCEWithLogitsLoss()
learn.metrics=[accuracy_thresh(0.5)]
In [ ]:
learn.load('0')
In [ ]:
learn.lr_find()
learn.sched.plot()
A Jupyter Widget
85%|████████▌ | 218/255 [02:12<00:22, 1.64it/s, loss=8.91]
In [ ]:
lr=4e-2
In [ ]:
learn.fit(lr,1,cycle_len=5,use_clr=(20,5))
A Jupyter Widget
epoch trn_loss val_loss <lambda>
0 0.02178 0.020653 0.991708
1 0.017927 0.020653 0.990241
2 0.015958 0.016115 0.993394
3 0.015172 0.015143 0.993696
4 0.014315 0.014679 0.99388
Out[ ]:
[0.014679321, 0.99388032489352751]
In [ ]:
learn.save('tmp')
In [ ]:
learn.load('tmp')
In [ ]:
learn.unfreeze()
learn.bn_freeze(True)
In [ ]:
lrs = np.array([lr/100,lr/10,lr])/4
In [ ]:
learn.fit(lrs,1,cycle_len=8,use_clr=(20,8))
A Jupyter Widget
epoch trn_loss val_loss mask_acc
0 0.038687 0.018685 0.992782
1 0.024906 0.014355 0.994933
2 0.025055 0.014737 0.995526
3 0.024155 0.014083 0.995708
4 0.013446 0.010564 0.996166
5 0.01607 0.010555 0.996096
6 0.019197 0.010883 0.99621
7 0.016157 0.00998 0.996393
Out[ ]:
[0.0099797687, 0.99639255659920833]
In [ ]:
learn.save('512')
In [ ]:
x,y = next(iter(md.val_dl))
py = to_np(learn.model(V(x)))
In [ ]:
ax = show_img(denorm(x)[0])
show_img(py[0]>0, ax=ax, alpha=0.5);
In [ ]:
ax = show_img(denorm(x)[0])
show_img(y[0], ax=ax, alpha=0.5);
1024x1024¶
In [ ]:
sz = 1024
bs = 4
In [ ]:
tfms = tfms_from_model(resnet34, sz, crop_type=CropType.NO, tfm_y=TfmType.CLASS, aug_tfms=aug_tfms)
datasets = ImageData.get_ds(MatchedFilesDataset, (trn_x,trn_y), (val_x,val_y), tfms, path=PATH)
md = ImageData(PATH, datasets, bs, num_workers=8, classes=None)
In [ ]:
denorm = md.trn_ds.denorm
x,y = next(iter(md.aug_dl))
x = denorm(x)
y = to_np(y)
In [ ]:
fig, axes = plt.subplots(2, 2, figsize=(8, 8))
for i,ax in enumerate(axes.flat):
show_img(x[i], ax=ax)
show_img(y[i], ax=ax, alpha=0.5)
plt.tight_layout(pad=0.1)
In [ ]:
simple_up = nn.Sequential(
nn.ReLU(),
StdUpsample(512,256),
StdUpsample(256,256),
StdUpsample(256,256),
StdUpsample(256,256),
nn.ConvTranspose2d(256, 1, 2, stride=2),
flatten_channel,
)
In [ ]:
models = ConvnetBuilder(resnet34, 0, 0, 0, custom_head=simple_up)
learn = ConvLearner(md, models)
learn.opt_fn=optim.Adam
learn.crit=nn.BCEWithLogitsLoss()
learn.metrics=[accuracy_thresh(0.5)]
In [ ]:
learn.load('512')
In [ ]:
learn.lr_find()
learn.sched.plot()
A Jupyter Widget
85%|████████▌ | 218/255 [02:12<00:22, 1.64it/s, loss=8.91]
In [ ]:
lr=4e-2
In [ ]:
learn.fit(lr,1,cycle_len=2,use_clr=(20,4))
A Jupyter Widget
epoch trn_loss val_loss <lambda>
0 0.01066 0.011119 0.996227
1 0.009357 0.009696 0.996553
Out[ ]:
[0.0096957013, 0.99655332546385511]
In [ ]:
learn.save('tmp')
In [ ]:
learn.load('tmp')
In [ ]:
learn.unfreeze()
learn.bn_freeze(True)
In [ ]:
lrs = np.array([lr/100,lr/10,lr])/8
In [ ]:
learn.fit(lrs,1,cycle_len=40,use_clr=(20,10))
A Jupyter Widget
epoch trn_loss val_loss mask_acc
0 0.015565 0.007449 0.997661
1 0.01979 0.008376 0.997542
2 0.014874 0.007826 0.997736
3 0.016104 0.007854 0.997347
4 0.023386 0.009745 0.997218
5 0.018972 0.008453 0.997588
6 0.013184 0.007612 0.997588
7 0.010686 0.006775 0.997688
8 0.0293 0.015299 0.995782
9 0.018713 0.00763 0.997638
10 0.015432 0.006575 0.9978
11 0.110205 0.060062 0.979043
12 0.014374 0.007753 0.997451
13 0.022286 0.010282 0.997587
14 0.015645 0.00739 0.997776
15 0.013821 0.00692 0.997869
16 0.022389 0.008632 0.997696
17 0.014607 0.00677 0.997837
18 0.018748 0.008194 0.997657
19 0.016447 0.007237 0.997899
20 0.023596 0.008211 0.997918
21 0.015721 0.00674 0.997848
22 0.01572 0.006415 0.998006
23 0.019519 0.007591 0.997876
24 0.011159 0.005998 0.998053
25 0.010291 0.005806 0.998012
26 0.010893 0.005755 0.998046
27 0.014534 0.006313 0.997901
28 0.020971 0.006855 0.998018
29 0.014074 0.006107 0.998053
30 0.01782 0.006561 0.998114
31 0.01742 0.006414 0.997942
32 0.016829 0.006514 0.9981
33 0.013148 0.005819 0.998033
34 0.023495 0.006261 0.997856
35 0.010931 0.005516 0.99812
36 0.015798 0.006176 0.998126
37 0.021636 0.005931 0.998067
38 0.012133 0.005496 0.998158
39 0.012562 0.005678 0.998172
Out[ ]:
[0.0056782686, 0.99817223208291195]
In [ ]:
learn.save('1024')
In [ ]:
x,y = next(iter(md.val_dl))
py = to_np(learn.model(V(x)))
In [ ]:
ax = show_img(denorm(x)[0])
show_img(py[0][0]>0, ax=ax, alpha=0.5);
In [ ]:
ax = show_img(denorm(x)[0])
show_img(y[0,...,-1], ax=ax, alpha=0.5);
In [ ]:
show_img(py[0][0]>0);
In [ ]:
show_img(y[0,...,-1]);
Fin¶
In [ ]: