forked from aharley/pips
-
Notifications
You must be signed in to change notification settings - Fork 0
/
test_on_crohd.py
350 lines (282 loc) · 12.7 KB
/
test_on_crohd.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
import time
import numpy as np
import timeit
import saverloader
from nets.raftnet import Raftnet
from nets.pips import Pips
import random
from utils.basic import print_, print_stats
import torch
import torch.nn as nn
from torch.utils.data import Dataset, DataLoader
from tensorboardX import SummaryWriter
import torch.nn.functional as F
from crohddataset import CrohdDataset
import utils.basic
import utils.improc
import utils.test
from fire import Fire
device = 'cuda'
random.seed(125)
np.random.seed(125)
def prep_sample(sample, N_max, S_stride=3, req_occlusion=False):
rgbs = sample['rgbs'].permute(0,1,4,2,S_stride).float()[:,::S_stride] # (1, S, C, H, W) in 0-255
boxlist = sample['boxlist'][0].float()[::S_stride] # (S, N, 4), N = n heads
xylist = sample['xylist'][0].float()[::S_stride] # (S, N, 2)
scorelist = sample['scorelist'][0].float()[::S_stride] # (S, N)
vislist = sample['vislist'][0].float()[::S_stride] # (S, N)
S, N, _ = xylist.shape
# collect valid heads
scorelist_sum = scorelist.sum(0) # (N)
seq_present = scorelist_sum == S
motion = torch.sqrt(torch.sum((xylist[1:] - xylist[:1])**2, dim=2)).sum(0) # (N)
seq_moving = motion > 150
seq_vis_init = vislist[:2].sum(0) == 2
seq_occlusion = vislist.sum(0) < 8
seq_visible = vislist.sum(0) == 8
if req_occlusion:
seq_valid = seq_present * seq_vis_init * seq_moving * seq_occlusion
else:
seq_valid = seq_present * seq_vis_init * seq_moving * seq_visible
if seq_valid.sum() == 0:
return None, True
kp_xys = xylist[:, seq_valid> 0].unsqueeze(0)
vis = vislist[:, seq_valid > 0].unsqueeze(0)
N = kp_xys.shape[2]
# print('N', N)
if N > N_max:
kp_xys = kp_xys[:,:,:N_max]
vis = vis[:,:,:N_max]
d = {
'rgbs': rgbs, # B, S, C, H, W
'trajs_g': kp_xys, # B, S, 2
'vis_g': vis, # B, S
}
return d, False
def run_dino(dino, d, sw):
rgbs = d['rgbs'].cuda()
trajs_g = d['trajs_g'].cuda() # B,S,N,2
vis_g = d['vis_g'].cuda() # B,S,N
valids = torch.ones_like(vis_g) # B,S,N
B, S, C, H, W = rgbs.shape
B, S1, N, D = trajs_g.shape
rgbs_ = rgbs.reshape(B*S, C, H, W)
H_, W_ = 512, 768
sy = H_/H
sx = W_/W
rgbs_ = F.interpolate(rgbs_, (H_, W_), mode='bilinear')
H, W = H_, W_
rgbs = rgbs_.reshape(B, S, C, H, W)
trajs_g[:,:,:,0] *= sx
trajs_g[:,:,:,1] *= sy
_, S, C, H, W = rgbs.shape
trajs_e = utils.test.get_dino_output(dino, rgbs, trajs_g, vis_g)
ate = torch.norm(trajs_e - trajs_g, dim=-1) # B, S, N
ate_all = utils.basic.reduce_masked_mean(ate, valids)
ate_vis = utils.basic.reduce_masked_mean(ate, valids*vis_g)
ate_occ = utils.basic.reduce_masked_mean(ate, valids*(1.0-vis_g))
metrics = {
'ate_all': ate_all.item(),
'ate_vis': ate_vis.item(),
'ate_occ': ate_occ.item(),
}
if sw is not None and sw.save_this:
sw.summ_traj2ds_on_rgbs('inputs_0/orig_trajs_on_rgbs', trajs_g, utils.improc.preprocess_color(rgbs), cmap='winter', linewidth=2)
sw.summ_traj2ds_on_rgbs('outputs/trajs_on_rgbs', trajs_e[0:1], utils.improc.preprocess_color(rgbs[0:1]), cmap='spring', linewidth=2)
gt_rgb = utils.improc.preprocess_color(sw.summ_traj2ds_on_rgb('inputs_0_all/single_trajs_on_rgb', trajs_g[0:1], torch.mean(utils.improc.preprocess_color(rgbs[0:1]), dim=1), cmap='winter', frame_id=metrics['ate_all'], only_return=True, linewidth=2))
gt_black = utils.improc.preprocess_color(sw.summ_traj2ds_on_rgb('inputs_0_all/single_trajs_on_rgb', trajs_g[0:1], torch.ones_like(rgbs[0:1,0])*-0.5, cmap='winter', frame_id=metrics['ate_all'], only_return=True, linewidth=2))
sw.summ_traj2ds_on_rgb('outputs/single_trajs_on_gt_rgb', trajs_e[0:1], gt_rgb[0:1], cmap='spring', linewidth=2)
sw.summ_traj2ds_on_rgb('outputs/single_trajs_on_gt_black', trajs_e[0:1], gt_black[0:1], cmap='spring', linewidth=2)
return metrics
def run_pips(model, d, sw):
rgbs = d['rgbs'].cuda()
trajs_g = d['trajs_g'].cuda() # B,S,N,2
vis_g = d['vis_g'].cuda() # B,S,N
valids = torch.ones_like(vis_g) # B,S,N
B, S, C, H, W = rgbs.shape
B, S1, N, D = trajs_g.shape
rgbs_ = rgbs.reshape(B*S, C, H, W)
H_, W_ = 768, 1280
sy = H_/H
sx = W_/W
rgbs_ = F.interpolate(rgbs_, (H_, W_), mode='bilinear')
H, W = H_, W_
rgbs = rgbs_.reshape(B, S, C, H, W)
trajs_g[:,:,:,0] *= sx
trajs_g[:,:,:,1] *= sy
_, S, C, H, W = rgbs.shape
preds, preds_anim, vis_e, stats = model(trajs_g[:,0], rgbs, iters=6, trajs_g=trajs_g, vis_g=vis_g, valids=valids, sw=sw)
ate = torch.norm(preds[-1] - trajs_g, dim=-1) # B, S, N
ate_all = utils.basic.reduce_masked_mean(ate, valids)
ate_vis = utils.basic.reduce_masked_mean(ate, valids*vis_g)
ate_occ = utils.basic.reduce_masked_mean(ate, valids*(1.0-vis_g))
metrics = {
'ate_all': ate_all.item(),
'ate_vis': ate_vis.item(),
'ate_occ': ate_occ.item(),
}
trajs_e = preds[-1]
if sw is not None and sw.save_this:
sw.summ_traj2ds_on_rgbs('inputs_0/orig_trajs_on_rgbs', trajs_g, utils.improc.preprocess_color(rgbs), cmap='winter', linewidth=2)
sw.summ_traj2ds_on_rgbs('outputs/trajs_on_rgbs', trajs_e[0:1], utils.improc.preprocess_color(rgbs[0:1]), cmap='spring', linewidth=2)
gt_rgb = utils.improc.preprocess_color(sw.summ_traj2ds_on_rgb('inputs_0_all/single_trajs_on_rgb', trajs_g[0:1], torch.mean(utils.improc.preprocess_color(rgbs[0:1]), dim=1), cmap='winter', frame_id=metrics['ate_all'], only_return=True, linewidth=2))
gt_black = utils.improc.preprocess_color(sw.summ_traj2ds_on_rgb('inputs_0_all/single_trajs_on_rgb', trajs_g[0:1], torch.ones_like(rgbs[0:1,0])*-0.5, cmap='winter', frame_id=metrics['ate_all'], only_return=True, linewidth=2))
sw.summ_traj2ds_on_rgb('outputs/single_trajs_on_gt_rgb', trajs_e[0:1], gt_rgb[0:1], cmap='spring', linewidth=2)
sw.summ_traj2ds_on_rgb('outputs/single_trajs_on_gt_black', trajs_e[0:1], gt_black[0:1], cmap='spring', linewidth=2)
return metrics
def run_raft(raft, d, sw):
rgbs = d['rgbs'].cuda()
trajs_g = d['trajs_g'].cuda() # B,S,N,2
vis_g = d['vis_g'].cuda() # B,S,N
valids = torch.ones_like(vis_g) # B,S,N
B, S, C, H, W = rgbs.shape
B, S1, N, D = trajs_g.shape
rgbs_ = rgbs.reshape(B*S, C, H, W)
H_, W_ = 768, 1280
sy = H_/H
sx = W_/W
rgbs_ = F.interpolate(rgbs_, (H_, W_), mode='bilinear')
H, W = H_, W_
rgbs = rgbs_.reshape(B, S, C, H, W)
trajs_g[:,:,:,0] *= sx
trajs_g[:,:,:,1] *= sy
_, S, C, H, W = rgbs.shape
prep_rgbs = utils.improc.preprocess_color(rgbs)
flows_e = []
for s in range(S-1):
rgb0 = prep_rgbs[:,s]
rgb1 = prep_rgbs[:,s+1]
flow, _ = raft(rgb0, rgb1, iters=32)
flows_e.append(flow)
flows_e = torch.stack(flows_e, dim=1) # B, S-1, 2, H, W
coords = []
coord0 = trajs_g[:,0] # B, N, 2
coords.append(coord0)
coord = coord0.clone()
for s in range(S-1):
delta = utils.samp.bilinear_sample2d(
flows_e[:,s], coord[:,:,0], coord[:,:,1]).permute(0,2,1) # B, N, 2, forward flow at the discrete points
coord = coord + delta
coords.append(coord)
trajs_e = torch.stack(coords, dim=1) # B, S, N, 2
ate = torch.norm(trajs_e - trajs_g, dim=-1) # B, S, N
ate_all = utils.basic.reduce_masked_mean(ate, valids)
ate_vis = utils.basic.reduce_masked_mean(ate, valids*vis_g)
ate_occ = utils.basic.reduce_masked_mean(ate, valids*(1.0-vis_g))
metrics = {
'ate_all': ate_all.item(),
'ate_vis': ate_vis.item(),
'ate_occ': ate_occ.item(),
}
if sw is not None and sw.save_this:
sw.summ_traj2ds_on_rgbs('inputs_0/orig_trajs_on_rgbs', trajs_g, utils.improc.preprocess_color(rgbs), cmap='winter', linewidth=2)
sw.summ_traj2ds_on_rgbs('outputs/trajs_on_rgbs', trajs_e[0:1], utils.improc.preprocess_color(rgbs[0:1]), cmap='spring', linewidth=2)
gt_rgb = utils.improc.preprocess_color(sw.summ_traj2ds_on_rgb('inputs_0_all/single_trajs_on_rgb', trajs_g[0:1], torch.mean(utils.improc.preprocess_color(rgbs[0:1]), dim=1), cmap='winter', frame_id=metrics['ate_all'], only_return=True, linewidth=2))
gt_black = utils.improc.preprocess_color(sw.summ_traj2ds_on_rgb('inputs_0_all/single_trajs_on_rgb', trajs_g[0:1], torch.ones_like(rgbs[0:1,0])*-0.5, cmap='winter', frame_id=metrics['ate_all'], only_return=True, linewidth=2))
sw.summ_traj2ds_on_rgb('outputs/single_trajs_on_gt_rgb', trajs_e[0:1], gt_rgb[0:1], cmap='spring', linewidth=2)
sw.summ_traj2ds_on_rgb('outputs/single_trajs_on_gt_black', trajs_e[0:1], gt_black[0:1], cmap='spring', linewidth=2)
return metrics
def main(
exp_name='crohd',
B=1,
S=8,
N=16,
modeltype='pips',
init_dir='reference_model',
req_occlusion=True,
stride=4,
log_dir='logs_test_on_crohd',
max_iters=0, # auto-select based on dataset
log_freq=100,
shuffle=False,
subset='all',
use_augs=False,
):
# the idea in this file is to evaluate on head tracking in croHD
assert(modeltype=='pips' or modeltype=='raft' or modeltype=='dino')
S_stride = 3 # subsample the frames this much
## autogen a name
model_name = "%d_%d_%d_%s" % (B, S, N, modeltype)
if req_occlusion:
model_name += "_occ"
else:
model_name += "_vis"
model_name += "_%s" % exp_name
import datetime
model_date = datetime.datetime.now().strftime('%H:%M:%S')
model_name = model_name + '_' + model_date
print('model_name', model_name)
writer_t = SummaryWriter(log_dir + '/' + model_name + '/t', max_queue=10, flush_secs=60)
dataset = CrohdDataset(seqlen=S*S_stride)
test_dataloader = DataLoader(
dataset,
batch_size=B,
shuffle=shuffle,
num_workers=12)
test_iterloader = iter(test_dataloader)
global_step = 0
if modeltype=='pips':
model = Pips(S=S, stride=stride).cuda()
_ = saverloader.load(init_dir, model)
model.eval()
elif modeltype=='raft':
model = Raftnet(ckpt_name='../RAFT/models/raft-things.pth').cuda()
model.eval()
elif modeltype=='dino':
patch_size = 8
model = torch.hub.load('facebookresearch/dino:main', 'dino_vits%d' % patch_size).cuda()
model.eval()
else:
assert(False) # need to choose a valid modeltype
n_pool = 10000
ate_all_pool_t = utils.misc.SimplePool(n_pool, version='np')
ate_vis_pool_t = utils.misc.SimplePool(n_pool, version='np')
ate_occ_pool_t = utils.misc.SimplePool(n_pool, version='np')
if max_iters==0:
max_iters = len(test_dataloader)
print('setting max_iters', max_iters)
while global_step < max_iters:
read_start_time = time.time()
global_step += 1
sw_t = utils.improc.Summ_writer(
writer=writer_t,
global_step=global_step,
log_freq=log_freq,
fps=5,
scalar_freq=int(log_freq/2),
just_gif=True)
returned_early = True
while returned_early:
try:
sample = next(test_iterloader)
except StopIteration:
test_iterloader = iter(test_dataloader)
sample = next(test_iterloader)
sample, returned_early = prep_sample(sample, N, S_stride, req_occlusion)
read_time = time.time()-read_start_time
iter_start_time = time.time()
with torch.no_grad():
if modeltype=='pips':
metrics = run_pips(model, sample, sw_t)
elif modeltype=='raft':
metrics = run_raft(model, sample, sw_t)
elif modeltype=='dino':
metrics = run_dino(model, sample, sw_t)
else:
assert(False) # need to choose a valid modeltype
if metrics['ate_all'] > 0:
ate_all_pool_t.update([metrics['ate_all']])
if metrics['ate_vis'] > 0:
ate_vis_pool_t.update([metrics['ate_vis']])
if metrics['ate_occ'] > 0:
ate_occ_pool_t.update([metrics['ate_occ']])
sw_t.summ_scalar('pooled/ate_all', ate_all_pool_t.mean())
sw_t.summ_scalar('pooled/ate_vis', ate_vis_pool_t.mean())
sw_t.summ_scalar('pooled/ate_occ', ate_occ_pool_t.mean())
iter_time = time.time()-iter_start_time
print('%s; step %06d/%d; rtime %.2f; itime %.2f; ate = %.2f; ate_pooled = %.2f' % (
model_name, global_step, max_iters, read_time, iter_time,
metrics['ate_all'], ate_all_pool_t.mean()))
writer_t.close()
if __name__ == '__main__':
Fire(main)