-
Notifications
You must be signed in to change notification settings - Fork 9
/
detect_room.py
437 lines (347 loc) · 12.8 KB
/
detect_room.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
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
"""
Get attributes about images
Inspired by https://github.com/CSAILVision/places365/blob/master/run_placesCNN_unified.py
"""
from pathlib import Path
import os
from typing import List, Iterator, Tuple, Optional, Union, Dict
import hashlib
import json
from multiprocessing import Pool
import urllib.request
import sys
import csv
from tqdm.auto import tqdm
import torch
from torchvision import transforms as trn
from torch import nn
from torch.utils.data._utils.collate import default_collate
from torch.nn import functional as F
import numpy as np
import cv2
from PIL import Image
import tap
from torch.utils.data import Dataset, DataLoader
import wideresnet as wideresnet
csv.field_size_limit(sys.maxsize)
TSV_FIELDNAMES = [
"listing_id",
"photo_id",
"category",
"attributes",
"is_indoor",
]
class Arguments(tap.Tap):
output: Path = Path("places365/detect.tsv")
images: Path = Path("images")
batch_size: int = 100
visualize: bool = False
num_cat: int = 5
num_attr: int = 10
num_splits: int = 1
start: int = 0
num_workers: int = 0
# hacky way to deal with the Pytorch 1.0 update
def recursion_change_bn(module: nn.Module) -> nn.Module:
if isinstance(module, nn.BatchNorm2d):
module.track_running_stats = 1 # type: ignore
else:
for i, (name, module1) in enumerate(module._modules.items()): # type: ignore
module1 = recursion_change_bn(module1)
return module
def download_url(url, cache_dir):
stem = hashlib.sha1(str(url).encode())
filename = cache_dir / stem.hexdigest()
if not filename.is_file():
urllib.request.urlretrieve(url, filename)
return filename
def load_labels(
cache_dir: Union[Path, str]
) -> Tuple[Tuple[str, ...], np.ndarray, List[str], np.ndarray]:
"""
prepare all the labels
"""
# indoor and outdoor relevant
filename_io = download_url(
"https://raw.githubusercontent.com/csailvision/places365/master/IO_places365.txt",
cache_dir,
)
with open(filename_io) as f:
lines = f.readlines()
labels_IO = []
for line in lines:
items = line.rstrip().split()
labels_IO.append(int(items[-1]) - 1) # 0 is indoor, 1 is outdoor
labels_IO = np.array(labels_IO)
# scene category relevant
filename_category = download_url(
"https://raw.githubusercontent.com/csailvision/places365/master/categories_places365.txt",
cache_dir,
)
_classes = list()
with open(filename_category) as class_file:
for line in class_file:
_classes.append(line.strip().split(" ")[0][3:])
classes = tuple(_classes)
# scene attribute relevant
filename_attribute = download_url(
"https://raw.githubusercontent.com/csailvision/places365/master/labels_sunattribute.txt",
cache_dir,
)
with open(filename_attribute) as f:
lines = f.readlines()
labels_attribute = [item.rstrip() for item in lines]
filename_W = download_url(
"http://places2.csail.mit.edu/models_places365/W_sceneattribute_wideresnet18.npy",
cache_dir,
)
W_attribute = np.load(filename_W)
return classes, labels_IO, labels_attribute, W_attribute
def get_tf():
# load the image transformer
tf = trn.Compose(
[
trn.Resize((224, 224)),
trn.ToTensor(),
trn.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]),
]
)
return tf
class NormalizeInverse(trn.Normalize):
"""
Undoes the normalization and returns the reconstructed images in the input domain.
"""
def __init__(self, mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]):
mean = torch.tensor(mean)
std = torch.tensor(std)
std_inv = 1 / (std + 1e-7) # type: ignore
mean_inv = -mean * std_inv
super().__init__(mean=mean_inv, std=std_inv)
def __call__(self, array: np.ndarray):
tensor = torch.tensor(array)
tensor = super().__call__(tensor.clone())
array = np.transpose(np.uint8(255 * tensor.numpy()), (1, 2, 0))
return array
class Hooker:
def __init__(self, model: nn.Module, features_names=("layer4", "avgpool")):
self.features: List[np.ndarray] = []
# this is the last conv layer of the resnet
for name in features_names:
model._modules.get(name).register_forward_hook(self) # type: ignore
def __call__(self, module: nn.Module, input, output):
self.features.append(output.data.cpu().numpy())
def reset(self):
self.features = []
# load the model
def load_model(cache_dir: Union[Path, str]) -> nn.Module:
# this model has a last conv feature map as 14x14
model_file = download_url(
"http://places2.csail.mit.edu/models_places365/wideresnet18_places365.pth.tar",
cache_dir,
)
model = wideresnet.resnet18(num_classes=365)
checkpoint = torch.load(model_file, map_location=lambda storage, loc: storage)
state_dict = {
str.replace(k, "module.", ""): v for k, v in checkpoint["state_dict"].items()
}
model.load_state_dict(state_dict)
# hacky way to deal with the upgraded batchnorm2D and avgpool layers...
for i, (name, module) in enumerate(model._modules.items()): # type: ignore
module = recursion_change_bn(model) # type: ignore
model.avgpool = torch.nn.AvgPool2d(kernel_size=14, stride=1, padding=0) # type: ignore
model.eval()
return model
def search_locations(image_folder: Path) -> List[Path]:
return [f for f in image_folder.iterdir() if f.is_dir()]
def load_photo_paths(locations: List[Path]) -> Iterator[Path]:
for location in tqdm(locations):
for photo in location.glob("*.jpg"):
yield photo
def load_photos(images: Path, cache_dir: Union[Path, str]) -> List[Union[str, Path]]:
photo_cache = Path(cache_dir) / "photos.txt"
if photo_cache.is_file():
with open(photo_cache, "r") as fid:
photos: List[Union[str, Path]] = [l.strip() for l in fid.readlines()]
else:
print("Preloading every images")
photos = list(images.rglob("*.jpg"))
with open(photo_cache, "w") as fid:
fid.writelines(f"{l}\n" for l in photos)
return photos
class ImageDataset(Dataset):
def __init__(self, photos: List[Union[Path, str]]):
self.photos = photos
self.tf = get_tf() # image transformer
def __len__(self):
return len(self.photos)
def __getitem__(
self, index: int
) -> Optional[Tuple[torch.Tensor, torch.Tensor, torch.Tensor]]:
path = Path(self.photos[index])
try:
image = Image.open(path)
image = image.convert("RGB")
except:
return None
tensor = self.tf(image)
listing_id, photo_id = map(int, path.stem.split("-"))
return torch.tensor(listing_id), torch.tensor(photo_id), tensor
def collate_fn(batch: Tuple):
batch = tuple([b for b in batch if b is not None])
if not batch:
return None
return default_collate(batch)
def class_activation_map(
feature_conv: np.ndarray, weight_softmax: np.ndarray, class_idx: List[int]
):
# generate the class activation maps upsample to 256x256
size_upsample = (256, 256)
nc, h, w = feature_conv.shape
output_cam = []
for _ in class_idx:
cam = weight_softmax[class_idx].dot(feature_conv.reshape((nc, h * w)))
cam = cam.reshape(h, w)
cam = cam - np.min(cam)
cam_img = cam / np.max(cam)
cam_img = np.uint8(255 * cam_img)
output_cam.append(cv2.resize(cam_img, size_upsample)) # type: ignore
return output_cam
def get_key(listing_id, photo_id) -> str:
return f"{listing_id}_{photo_id}"
def is_indoor(idx, labels_io):
# vote for the indoor or outdoor
io_image = np.mean(labels_io[idx[:10]])
ans = bool(io_image < 0.5)
return io_image, ans
def softmax(x):
"""Compute softmax values for each sets of scores in x."""
e_x = np.exp(x - np.max(x))
return e_x / e_x.sum(axis=0) # type: ignore
@torch.no_grad()
def run_model(
batch: List[torch.Tensor],
model,
hook,
classes: Tuple[str, ...],
labels_IO: np.ndarray,
labels_attribute: List[str],
W_attribute: np.ndarray,
num_cat: int,
num_attr: int,
weight_softmax: Optional[np.ndarray] = None,
) -> List[Dict]:
listing_ids, photo_ids, input_img = batch
# forward pass
logit = model.forward(input_img.cuda())
h_x = F.softmax(logit, 1)
detections = []
for i, p in enumerate(h_x): # type: ignore
listing_id = int(listing_ids[i])
photo_id = int(photo_ids[i])
key = get_key(listing_id, photo_id)
probs, idx = p.sort(0, True) # type: ignore
probs = probs.detach().cpu().numpy()
idx = idx.detach().cpu().numpy()
# scene category
category = [(probs[j], classes[idx[j]]) for j in range(0, num_cat)]
# output the scene attributes
ft = [np.squeeze(f[i]) for f in hook.features]
responses_attribute = softmax(W_attribute.dot(ft[1]))
idx_a = np.argsort(responses_attribute)
attributes = [
(responses_attribute[idx_a[j]], labels_attribute[idx_a[j]])
for j in range(-1, -num_attr, -1)
]
detections.append(
{
"listing_id": listing_id,
"photo_id": photo_id,
"category": category,
"attributes": attributes,
"is_indoor": is_indoor(idx, labels_IO),
}
)
# generate class activation mapping
if weight_softmax is not None:
ca_map = class_activation_map(ft[0], weight_softmax, [idx[0]])[0]
# render the CAM and output
img = NormalizeInverse()(input_img[i])
height, width, _ = img.shape # type: ignore
heatmap = cv2.applyColorMap( # type: ignore
cv2.resize(ca_map, (width, height)), cv2.COLORMAP_JET # type: ignore
)
result = heatmap * 0.4 + img * 0.5 # type: ignore
cv2.imwrite(f"examples/{key}-heatmap.jpg", result) # type: ignore
cv2.imwrite(f"examples/{key}-image.jpg", img[:, :, ::-1]) # type: ignore
hook.reset()
return detections
class NumpyEncoder(json.JSONEncoder):
"""Special json encoder for numpy types"""
def default(self, obj):
if isinstance(obj, np.integer):
return int(obj)
elif isinstance(obj, np.floating):
return float(obj)
elif isinstance(obj, np.ndarray):
return obj.tolist()
return json.JSONEncoder.default(self, obj)
def save_json(data, filename: Union[str, Path]):
with open(filename, "w") as fid:
json.dump(data, fid, indent=2, cls=NumpyEncoder)
def detection(args: Arguments, proc_id: int, cache_dir: Union[Path, str]):
# load the labels
classes, labels_IO, labels_attribute, W_attribute = load_labels(cache_dir)
model = load_model(cache_dir)
hook = Hooker(model)
# load the transformer
# get the softmax weight
params = list(model.parameters())
if args.visualize:
weight_softmax = params[-2].data.numpy()
weight_softmax[weight_softmax < 0] = 0
else:
weight_softmax = None
photos = load_photos(args.images, cache_dir)
print("The dataset contains a total of", len(photos))
photos = photos[proc_id :: args.num_splits]
print(
"The split", proc_id, "over", args.num_splits, "contains", len(photos), "photos"
)
dataset = ImageDataset(photos)
dataloader = DataLoader(
dataset,
batch_size=args.batch_size,
num_workers=args.num_workers,
collate_fn=collate_fn, # type: ignore
)
model = model.cuda()
args.output.parent.mkdir(exist_ok=True, parents=True)
filename = args.output.parent / f"{args.output.stem}.{proc_id}.tsv"
print(f"Start split {proc_id} on {len(dataset)} photos")
with open(filename, "wt") as tsvfile:
writer = csv.DictWriter(tsvfile, delimiter="\t", fieldnames=TSV_FIELDNAMES)
for batch in tqdm(dataloader):
if batch is None:
continue
detections = run_model(
batch,
model,
hook,
classes,
labels_IO,
labels_attribute,
W_attribute,
num_cat=args.num_cat,
num_attr=args.num_attr,
weight_softmax=weight_softmax,
)
for d in detections:
writer.writerow(d)
if __name__ == "__main__":
args = Arguments().parse_args()
print(args)
cache_dir = Path.home() / ".cache" / args.output.parent.name
cache_dir.mkdir(exist_ok=True, parents=True)
local_rank = int(os.environ.get('LOCAL_RANK', 0))
start = max(local_rank, 0) + args.start
detection(args, start, cache_dir)