-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtest_with_flow.py
234 lines (175 loc) · 10.8 KB
/
test_with_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
import os
import torch
import torch.nn as nn
import torch.nn.functional as F
import numpy as np
import cv2
import argparse
import csv
from utils.image_process import normalize_image
from data_readers.video_readers import ImageReader
from e2v.e2v_model import *
from utils.configs import set_configs
from utils.data_io import ImageWriter, EvalWriter, FlowWriter, show_whole_img, show_flow
from utils.evaluate import mse, psnr, ssim, PerceptualLoss
from utils.flow_utils import FrameWarp
from loss import FlowReconLoss
print("Selected device:", torch.cuda.current_device())
class Reconstructor(nn.Module):
def __init__(self, cfgs, device):
super(Reconstructor, self).__init__()
self.image_dim = cfgs.image_dim
self.reader_type = cfgs.reader_type
self.model_mode = cfgs.model_mode
self.device = device
self.num_load_frames = cfgs.test_img_num
self.test_data_name = cfgs.test_data_name
self.limit_num_events = cfgs.num_events
self.test_data_mode = cfgs.test_data_mode
self.warp_mode = cfgs.warp_mode
self.path_to_sequences = []
for folder_name in os.listdir(cfgs.path_to_test_data):
if os.path.isdir(os.path.join(cfgs.path_to_test_data, folder_name)):
self.path_to_sequences.append(os.path.join(cfgs.path_to_test_data, folder_name))
self.path_to_sequences.sort()
self.video_renderer = ImageReader(cfgs, device=self.device)
# initialize reconstruction network
if self.model_mode == 'cista-eiflow':
self.model = DCEIFlowCistaNet(cfgs)
elif self.model_mode == 'cista-eraft':
self.model = ERAFTCistaNet(cfgs)
elif self.model_mode == 'cista-idnet':
self.model = IDCistaNet(cfgs)
else:
assert self.model_mode in ['cista-eiflow', 'cista-eraft']
# Load pretrained CISTA-Flow model
if cfgs.load_epoch_for_test:
self.model_name = cfgs.path_to_test_model.split('/')[-2]
cfgs.path_to_test_model = cfgs.path_to_test_model + '{}_{}.pth.tar'.format(self.model_name, cfgs.load_epoch_for_test)
self.model_name = self.model_name + '/{}'.format(cfgs.load_epoch_for_test)
else:
self.model_name = os.path.splitext(cfgs.path_to_test_model.split('/')[-1])[0]
checkpoint = torch.load(cfgs.path_to_test_model, map_location='cuda:0')
self.model.load_state_dict(checkpoint['state_dict'], strict=False) #True
# Replace CISTA-LSTC network with specified pretrained model
if cfgs.path_to_e2v:
checkpoint = torch.load(cfgs.path_to_e2v, map_location='cuda:0')
self.model.cista_net.load_state_dict(checkpoint['state_dict'], strict=True) #True
print(self.model)
print('Model name: ', self.model_name)
self.model.to(device)
self.model.eval()
total_params = sum(p.numel() for p in self.model.parameters())
print(f"Total parameters: {total_params}")
# for float32
total_memory_MB = total_params * 32 / 8 / 1024 / 1024
print(f"Estimated model memory size: {total_memory_MB:.2f} MB")
self.frame_warp = FrameWarp(mode=cfgs.warp_mode)
self.loss_fn = FlowReconLoss(cfgs.image_dim, self.frame_warp, ds=cfgs.ds, is_bi=False).to(device)
def forward(self):
# torch.backends.cudnn.enabled = False
with torch.no_grad():
all_seq_test_results, all_seq_flow_results = [], []
whole_test_mean, whole_flow_mean = [], []
num_total_frames, num_total_flow_frames = 0, 0
metric_keys = None
for seq_id, path_to_sequence_folder in enumerate(self.path_to_sequences):
dataset_name=path_to_sequence_folder.split('/')[-1].split('.')[0]
if self.test_data_name is not None and dataset_name != self.test_data_name:
continue
self.video_renderer.initialize(path_to_sequence_folder, self.num_load_frames)
states = None
prev_image = None
flow_states = None
image_writer = ImageWriter(cfgs, self.model_name, dataset_name)
eval_writer = EvalWriter(cfgs, self.model_name, dataset_name)
flow_writer = FlowWriter(cfgs, self.model_name, dataset_name)
all_test_results = []
all_flow_results = []
frame_idx = 0
prev_events = None
# gt_prev_frame = None
while not self.video_renderer.ending:
events, frame_pack, gt_frame, flow = self.video_renderer.update_event_frame_flow_pack(mode=self.test_data_mode)
if prev_image is None:
prev_image = torch.zeros([1, 1, self.image_dim[0], self.image_dim[1]], dtype=torch.float32, device=self.device)
input_data = dict([])
input_gt = dict([])
for i, (evs, gt_prev_frame, gt_flow) in enumerate(zip(events, frame_pack, flow)):
evs = torch.unsqueeze(torch.from_numpy(evs), axis=0).to(self.device)
gt_prev_frame = torch.from_numpy(gt_prev_frame).unsqueeze(0).unsqueeze(0).to(self.device)
gt_flow = torch.from_numpy(gt_flow).unsqueeze(0).to(self.device)
gt_frame_tensor = frame_pack[i+1] if i < len(frame_pack)-1 else gt_frame
gt_frame_tensor = torch.from_numpy(gt_frame_tensor).unsqueeze(0).unsqueeze(0).to(self.device)
input_data['event_voxel'] = evs
input_data['rec_img0'] = prev_image
if cfgs.is_gt_flow:
input_gt['gt_flow'] = gt_flow
if self.model_mode in ['cista-eiflow']:
pred_image, batch_flow, states = self.model(input_data, states, input_gt)
elif self.model_mode in ['cista-eraft']:
if frame_idx == 0:
evs_old = torch.zeros_like(evs)
input_data['event_voxel_old'] = evs_old
pred_image, batch_flow, states = self.model(input_data, states, input_gt)
evs_old = evs.clone()
elif self.model_mode in ['cista-idnet']:
if frame_idx == 0:
flow_init = None
pred_image, batch_flow, states = self.model(input_data, states, flow_init, input_gt)
flow_init = batch_flow['next_flow']
prev_image = pred_image.clone()
# if cfgs.display_test:
# # show_whole_img(evs, init_flow, pred_flow) #torch.from_numpy(gt_frame).float().unsqueeze(0).unsqueeze(0))
# show_flow(evs, batch_flow['flow_final'], gt_flow, self.frame_warp.warp_frame(gt_prev_frame, batch_flow['flow_final'])-gt_frame_tensor, self.frame_warp.warp_frame(gt_prev_frame, gt_flow)-gt_frame_tensor) #self.backwarp(gt_prev_frame, gt_flow)-gt_frame_tensor,
batch_target = dict(
gt_img0 = gt_prev_frame,
gt_img1 = gt_frame_tensor,
gt_flow = gt_flow,
)
rec_metrics, flow_metrics = self.loss_fn.evaluate(pred_image, batch_flow['flow_final'], batch_target)
pred_image_uint8 = np.uint8(pred_image.squeeze().cpu().data.numpy()*255.)
image_writer(pred_image_uint8, frame_idx+1)
flow_writer(batch_flow['flow_final'].squeeze().cpu().data.numpy(), frame_idx)
if frame_idx >= 3:
metrics = rec_metrics | flow_metrics
if metric_keys is None:
metric_keys = list(metrics.keys())
all_test_results.append(list(metrics.values()))
frame_idx += 1
all_test_results = np.array(all_test_results)
mean_test_results = all_test_results.mean(0)
mean_results = [eval_writer.dataset_name] + list(np.array(mean_test_results).round(4)) + [len(all_test_results)]
all_seq_test_results.append(mean_results)
whole_test_mean.append(mean_test_results)
num_total_frames += len(all_test_results)
eval_results = ' '.join(['{}: {:.4f}, '.format(metric_keys[i], mean_test_results[i]) for i in range(len(metric_keys))])
print('\nTest set {}: Average results for {:d} frames: {} \n'.format(
eval_writer.dataset_name, len(all_test_results), eval_results))
name_results = ['Dataset'] + metric_keys + ['N_frames']
eval_writer(name_results, mean_results)
mean_all_test_results = np.array(whole_test_mean).mean(0)
eval_results = ' '.join(['{}: {:.4f}, '.format(metric_keys[i], mean_all_test_results[i]) for i in range(len(metric_keys))])
print('\n Average results for {:d} frames: {} \n'.format(
num_total_frames, eval_results))
name_results =['Dataset'] + metric_keys + ['N_frames']
all_seq_test_results.append(['mean'] + list(np.array(mean_all_test_results).round(4)) + [num_total_frames] )
if cfgs.test_data_name is None:
output_folder = eval_writer.output_data_folder.split('/')[:-1]
output_folder = '/'.join(cur_str for cur_str in output_folder)
if not os.path.exists(output_folder):
os.makedirs(output_folder)
path_to_write_csv = os.path.join(output_folder, 'all.csv')
with open(path_to_write_csv, 'a+', newline='') as f:
writer = csv.writer(f, delimiter='\t')
writer.writerow(name_results)
writer.writerows(all_seq_test_results)
f.close()
if __name__ == '__main__':
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
print('device: ', device)
parser = argparse.ArgumentParser()
set_configs(parser)
cfgs = parser.parse_args()
reconstuctor = Reconstructor(cfgs, device)
reconstuctor()