-
Notifications
You must be signed in to change notification settings - Fork 0
/
eval_flow.py
295 lines (249 loc) · 11.1 KB
/
eval_flow.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
import argparse
import collections
import os
import mlflow
import torch
from torch.optim import *
from configs.parser import YAMLParser
from dataloader.h5 import H5Loader
from loss.flow import EvalCriteria
from models.model import *
from utils.activity_packager import Packager
from utils.homography import HomographyWarping
from utils.iwe import compute_pol_iwe
from utils.utils import load_model, create_model_dir, save_state_dict
from utils.mlflow import log_config, log_results
from utils.visualization import Visualization
def test(args, config_parser):
mlflow.set_tracking_uri(args.path_mlflow)
run = mlflow.get_run(args.runid)
config = config_parser.merge_configs(run.data.params)
config = config_parser.combine_entries(config)
# configs
if config["loader"]["batch_size"] > 1:
config["vis"]["activity"] = False
config["vis"]["bars"] = False
config["vis"]["enabled"] = False
config["vis"]["ground_truth"] = False
config["vis"]["store"] = False
distortion = False
if "distortion" in config.keys():
if config["distortion"]["undistort"]:
distortion = True
# create directory for inference results
path_results = create_model_dir(args.path_results, args.runid)
# store validation settings
eval_id = log_config(path_results, args.runid, config)
# initialize settings
device = config_parser.device
kwargs = config_parser.loader_kwargs
config["loader"]["device"] = device
# visualization tool
vis = Visualization(config, eval_id=eval_id, path_results=path_results)
if config["vis"]["activity"]:
activity_dir = path_results + "activity/" + str(eval_id) + "/"
if not os.path.exists(activity_dir):
os.makedirs(activity_dir)
# data loader
data = H5Loader(config)
dataloader = torch.utils.data.DataLoader(
data,
drop_last=True,
batch_size=config["loader"]["batch_size"],
collate_fn=data.custom_collate,
worker_init_fn=config_parser.worker_init_fn,
**kwargs,
)
# activate axonal delays for loihi compatible networks
if config["model"]["name"] in ["LoihiRec4ptNet", "SplitLoihiRec4ptNet"]:
config["model"]["spiking_neuron"]["delay"] = True
# model initialization and settings
num_bins = 2 if config["data"]["voxel"] is None else config["data"]["voxel"]
model = eval(config["model"]["name"])(config["model"].copy(), config["data"]["crop"], num_bins)
model = model.to(device)
model = load_model(args.runid, model, device)
save_state_dict(args.runid, model)
model.eval()
if config["vis"]["activity"]:
model.store_activity()
# homogrpahy projection
homography = HomographyWarping(config, flow_scaling=config["loss"]["flow_scaling"], K=data.cam_mtx)
# validation metric
criteria = EvalCriteria(config, device)
# inference loop
val_results = {}
end_test = False
prev_sequence = None
axonal_delays = None
gt_bf = {}
ts_bf = None
event_list_bf = None
event_mask_bf = None
event_list_pol_mask_bf = None
with torch.no_grad():
while True:
for inputs in dataloader:
sequence = data.files[data.batch_idx[0] % len(data.files)].split("/")[-1].split(".")[0]
if config["vis"]["activity"]:
if not os.path.exists(activity_dir + sequence + ".h5"):
packager = Packager(activity_dir + sequence + ".h5")
sample_idx = 0
if data.new_seq:
data.new_seq = False
if config["vis"]["ground_truth"] and prev_sequence is not None:
if prev_sequence not in val_results.keys():
val_results[prev_sequence] = {}
val_results[prev_sequence]["AEE"] = criteria.aee(criteria.gt).cpu().numpy()
vis.store_ground_truth(criteria.gt, prev_sequence)
homography.reset_pose_gt()
ts_bf.clear()
event_list_bf.clear()
event_mask_bf.clear()
event_list_pol_mask_bf.clear()
for key in gt_bf.keys():
gt_bf[key].clear()
vis.final_spike_rate = None
model.reset_states()
criteria.reset_gt()
criteria.reset_rsat()
# finish inference loop
if data.seq_num >= len(data.files):
end_test = True
break
# forward pass
x = model(inputs["net_input"].to(device))
# axonal delays
if axonal_delays is None:
axonal_delays = getattr(model.encoder_unet, "delays", 0)
ts_bf = collections.deque(maxlen=axonal_delays + 1)
event_list_bf = collections.deque(maxlen=axonal_delays + 1)
event_mask_bf = collections.deque(maxlen=axonal_delays + 1)
event_list_pol_mask_bf = collections.deque(maxlen=axonal_delays + 1)
for key in inputs.keys():
if key.split("_")[0] == "gt":
gt_bf[key] = collections.deque(maxlen=axonal_delays + 1)
# input buffer
ts_bf.append(data.last_proc_timestamp)
event_list_bf.append(inputs["event_list"].to(device))
event_mask_bf.append(inputs["event_mask"].to(device))
event_list_pol_mask_bf.append(inputs["event_list_pol_mask"].to(device))
for key in gt_bf.keys():
gt_bf[key].append(inputs[key].to(device))
# homography projection
flow_vectors = x["flow_vectors"].clone()
if config["data"]["mode"] == "time":
flow_vectors *= config["data"]["window"] / 0.001 # (flow in px/ms -> flow in px/input_time)
flow_list = homography.get_flow_map(flow_vectors)
# optical flow scaling
x["flow_vectors"] *= config["loss"]["flow_scaling"]
# mask flow for visualization
flow_vis = flow_list[-1].clone()
flow_vis *= event_mask_bf[0]
# image of warped events
iwe = compute_pol_iwe(
flow_vis,
event_list_bf[0],
config["loader"]["resolution"],
event_list_pol_mask_bf[0][:, :, 0:1],
event_list_pol_mask_bf[0][:, :, 1:2],
round_idx=False,
distortion=distortion,
)
iwe_window_vis = None
events_window_vis = None
masked_window_flow_vis = None
# update criteria
if len(ts_bf) == axonal_delays + 1:
criteria.event_flow_association(
flow_list, event_list_bf[0], event_list_pol_mask_bf[0], event_mask_bf[0]
)
if config["vis"]["ground_truth"]:
gt_flow_list = homography.pose_to_4ptflow(gt_bf["gt_position_OT"][0], gt_bf["gt_Euler_imu"][0])
criteria.update_gt(ts_bf[0], x["flow_vectors"], gt_flow_list)
# validation
if criteria.num_passes >= config["data"]["passes_loss"]:
# compute metric
deblurring_metric = criteria.rsat()
# accumulate results
for batch in range(config["loader"]["batch_size"]):
if sequence not in val_results.keys():
val_results[sequence] = {}
val_results[sequence]["it"] = 0
val_results[sequence]["RSAT"] = 0
val_results[sequence]["Hamming"] = 0
if event_list_bf[0].shape[1] > 0:
val_results[sequence]["it"] += 1
val_results[sequence]["RSAT"] += deblurring_metric[batch].cpu().numpy()
# visualize
if (config["vis"]["enabled"] or config["vis"]["store"]) and criteria.num_passes > 1:
events_window_vis = criteria.compute_window_events()
iwe_window_vis = criteria.compute_window_iwe()
masked_window_flow_vis = criteria.compute_masked_window_flow()
# reset criteria
criteria.reset_rsat()
prev_sequence = sequence
# visualize
if config["vis"]["bars"]:
for bar in data.open_files_bar:
bar.next()
if config["vis"]["enabled"]:
vis.update(
inputs,
flow_vis,
iwe,
events_window_vis,
masked_window_flow_vis,
iwe_window_vis,
flow_vectors=x["flow_vectors"],
)
if config["vis"]["store"]:
vis.store(
inputs,
flow_vis,
sequence,
iwe,
events_window_vis,
masked_window_flow_vis,
iwe_window_vis,
flow_vectors=x["flow_vectors"],
vision_spikes=x["spikes"].clone(),
ts=data.last_proc_timestamp,
)
if config["vis"]["activity"]:
if "activity" in x.keys() and x["activity"] is not None:
for key in x["activity"]:
packager.package_array(x["activity"][key], sample_idx, dir=key)
packager.package_array(x["flow_vectors"], sample_idx, dir="4pt")
sample_idx += 1
if end_test:
break
if config["vis"]["bars"]:
for bar in data.open_files_bar:
bar.finish()
# store validation config and results
results = {}
results["AEE"] = {}
results["RSAT"] = {}
results["Hamming"] = {}
for key in val_results.keys():
results["RSAT"][key] = str(val_results[key]["RSAT"] / val_results[key]["it"])
if config["vis"]["ground_truth"]:
results["AEE"][key] = str(val_results[key]["AEE"])
log_results(args.runid, results, path_results, eval_id)
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument("runid", help="mlflow model run")
parser.add_argument(
"--config",
default="configs/eval_flow.yml",
help="config file, overwrites mlflow settings",
)
parser.add_argument(
"--path_mlflow",
default="",
help="location of the mlflow ui",
)
parser.add_argument("--path_results", default="results_inference/")
args = parser.parse_args()
# launch testing
test(args, YAMLParser(args.config))