-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathtest.py
254 lines (217 loc) · 11.9 KB
/
test.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
import os
import sys
import cv2
import math
import torch
import argparse
import numpy as np
from tqdm import tqdm
from torch.nn import functional as F
from model.pytorch_msssim import ssim_matlab
from model.trainer import Model
from dataset.DualRealDataset import *
from lpips import lpips
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
loss_fn_alex = lpips.LPIPS(net='alex').to(device)
parser = argparse.ArgumentParser()
parser.add_argument('--training', default=False, type=bool)
parser.add_argument('--input_num', default=2, type=int, help='input images number')
parser.add_argument('--input_dir', default='/media/zhongyi/D/data/GOPRO_RSGR', type=str, required=True, help='path to the input dataset folder')
parser.add_argument('--dataset_name', default='realBR', type=str, required=True, help='Name of dataset to be used')
parser.add_argument('--data_mode1', default='Blur', type=str, help='Mode of input data')
parser.add_argument('--data_mode2', default='RS', type=str, help='Mode of input data')
parser.add_argument('--temporal', action='store_true',default=False)
parser.add_argument('--batch_size', default=16, type=int, help='minibatch size')
parser.add_argument('--output_num', default=5, type=int, help='final output channel of the network')
parser.add_argument('--output_dir', default='', type=str, required=True, help='path to save testing output')
parser.add_argument('--model_dir', default="./train_log/GOPROBase_RSGR_3_5/best.ckpt",
type=str, help='path to the pretrained model folder')
parser.add_argument('--keep_frames', action='store_true', default=False, help='save interpolated frames')
parser.add_argument('--keep_flows', action='store_true', default=False, help='save predicted flows')
args = parser.parse_args()
def flow2rgb(flow_map_np):
h, w, _ = flow_map_np.shape
rgb_map = np.ones((h, w, 3)).astype(np.float32)
normalized_flow_map = flow_map_np / (np.abs(flow_map_np).max())
rgb_map[:, :, 0] += normalized_flow_map[:, :, 0]
rgb_map[:, :, 1] -= 0.5 * (normalized_flow_map[:, :, 0] + normalized_flow_map[:, :, 1])
rgb_map[:, :, 2] += normalized_flow_map[:, :, 1]
return rgb_map.clip(0, 1)
def test(model):
model.load_model(path=args.model_dir)
data_root = os.path.join(args.input_dir, args.dataset_name)
if args.dataset_name == 'realBR':
args.inter_num = 16
args.intra_num = 9
dataset_val = DualRealDataset(dataset_cls='test',\
input_num=args.input_num,\
output_num=args.output_num,\
data_root=data_root,\
data_mode1 = args.data_mode1,\
data_mode2 = args.data_mode2,\
inter_num = args.inter_num,\
intra_num = args.intra_num, temp=args.temporal)
elif args.dataset_name == 'GOPRO-VFI_copy':
if args.output_num >8:
raise Exception('Wrong output number!!!')
args.inter_num = 0
args.intra_num = 8
dataset_val = DualRealDataset(dataset_cls='test',\
input_num=args.input_num,\
output_num=args.output_num,\
data_root=data_root,\
data_mode1 = args.data_mode1,\
data_mode2 = args.data_mode2,\
inter_num = args.inter_num,\
intra_num = args.intra_num, temp=args.temporal)
else:
raise Exception('Not supported data!!!')
val_data = DataLoader(dataset_val, batch_size=args.batch_size, pin_memory=True, num_workers=8)
psnr_list = []
psnr_dict = {}
ssim_list = []
ssim_dict = {}
lpips_list = []
lpips_dict = {}
psnr_time = {}
ssim_time = {}
lpips_time = {}
for i, all_data in enumerate(tqdm(val_data)):
data = all_data[0]
img_ids = all_data[1]
gt_ids = np.array(all_data[2]).T
data_gpu = data.to(device, non_blocking=True) / 255.
imgs_tensor = data_gpu[:, :3*args.input_num]
gts_tensor = data_gpu[:, 3*args.input_num:]
##### temporal-order encoding
batch,_,height,width = imgs_tensor.shape
rs_encode = torch.arange(0,height).type_as(imgs_tensor).unsqueeze(1).repeat(1,width) ##(h,w)
latent_gs_encode = []
for out_i in range(0,args.output_num):
gs_encode = torch.Tensor([(height-1)//(args.output_num-1)*out_i]).type_as(imgs_tensor).unsqueeze(0).repeat(height,width) #(h,w)
latent_gs_encode.append(gs_encode)
latent_gs_encodes = torch.stack(latent_gs_encode,dim=0) ##(output_num,h,w)
latent_gs_encodes = rs_encode.unsqueeze(0) - latent_gs_encodes ##(output_num*1,h,w)
latent_gs_encodes = latent_gs_encodes.unsqueeze(0).repeat(batch,1,1,1) ##(batch,output_num*1,h,w)
with torch.no_grad():
preds, flow_list,warped_imgs_list = model.inference(imgs_tensor,latent_gs_encodes)
flows = flow_list[2] # (batch_size,2*output_num*2,h,w)
warped_imgs = warped_imgs_list[2]
batch_size = imgs_tensor.shape[0]
for b_id in range(batch_size):
pred = preds[b_id] # (3*output_num,h,w)
gt_tensor = gts_tensor[b_id] # (3*output_num,h,w)
flow = flows[b_id] # (2*output_num*2,h,w)
warped_img = warped_imgs[b_id] # (3*output_num*2,h,w)
gt_id = gt_ids[b_id]
img_id = img_ids[b_id]
seq_name = img_id.split('/')[1]
img_name = img_id.split('/')[3]
save_path = args.output_dir+'/'+args.dataset_name+'/'+seq_name+'/'+img_name
if args.keep_flows is True or args.keep_frames is True:
if not os.path.exists(save_path):
os.makedirs(save_path, exist_ok=True)
flow_t2b = flow[:2*args.output_num]
flow_b2t = flow[args.output_num*2:]
warped_img_t2b = warped_img[:3*args.output_num]
warped_img_b2t = warped_img[3*args.output_num:]
for o_id in range(args.output_num):
p_img = pred[3*o_id:3*(o_id+1)] # (3,h,w)
g_img = gt_tensor[3*o_id:3*(o_id+1)]# (3,h,w)
f_img_t2b = flow_t2b[2*o_id:2*(o_id+1)]# (2,h,w)
f_img_b2t = flow_b2t[2*o_id:2*(o_id+1)]# (2,h,w)
w_img_t2b = warped_img_t2b[3*o_id:3*(o_id+1)] # (3,h,w)
w_img_b2t = warped_img_b2t[3*o_id:3*(o_id+1)] # (3,h,w)
g_id = gt_id[o_id] # str
ssim = ssim_matlab(g_img.unsqueeze(0),p_img.unsqueeze(0)).cpu().numpy()
MAX_DIFF = 1
mse = torch.mean((g_img - p_img) * (g_img - p_img)).cpu().data
psnr = 10* math.log10( MAX_DIFF**2 / mse )
lpips=loss_fn_alex(p_img, g_img).cpu().item()
if args.keep_frames is True:
gt_name = g_id.split('/')[-1]+'.png'
p_img_s = (p_img.permute(1,2,0).cpu().numpy() * 255).astype('uint8')
cv2.imwrite(os.path.join(save_path,gt_name),p_img_s)
if args.keep_flows is True:
gt_name_t2b = g_id.split('/')[-1]+'_flow_t2b.png'
f_img_s_t2b = (flow2rgb(f_img_t2b.permute(1,2,0).cpu().numpy()[:,:,::-1])*255).astype('uint8')
cv2.imwrite(os.path.join(save_path,gt_name_t2b),f_img_s_t2b)
gt_name_b2t = g_id.split('/')[-1]+'_flow_b2t.png'
f_img_s_b2t = (flow2rgb(f_img_b2t.permute(1,2,0).cpu().numpy()[:,:,::-1])*255).astype('uint8')
cv2.imwrite(os.path.join(save_path,gt_name_b2t),f_img_s_b2t)
time_id = o_id+1
if time_id not in lpips_time:
lpips_time[time_id] = []
lpips_time[time_id].append(lpips)
if time_id not in ssim_time:
ssim_time[time_id] = []
ssim_time[time_id].append(ssim)
if time_id not in psnr_time:
psnr_time[time_id] = []
psnr_time[time_id].append(psnr)
if seq_name not in lpips_dict:
lpips_dict[seq_name]={}
if img_name not in lpips_dict[seq_name]:
lpips_dict[seq_name][img_name] = {}
lpips_dict[seq_name][img_name][g_id.split('/')[-1]]= format(lpips,'.4f')
lpips_list.append(lpips)
if seq_name not in psnr_dict:
psnr_dict[seq_name]={}
if img_name not in psnr_dict[seq_name]:
psnr_dict[seq_name][img_name] = {}
psnr_dict[seq_name][img_name][g_id.split('/')[-1]]=format(psnr,'.4f')
psnr_list.append(psnr)
if seq_name not in ssim_dict:
ssim_dict[seq_name]={}
if img_name not in ssim_dict[seq_name]:
ssim_dict[seq_name][img_name] = {}
ssim_dict[seq_name][img_name][g_id.split('/')[-1]]=format(ssim,'.4f')
ssim_list.append(ssim)
save_dir = args.output_dir+'/'+args.dataset_name
if not os.path.exists(save_dir):
os.makedirs(save_dir, exist_ok=True)
# keep txt record
for seq_name,img_dict in psnr_dict.items():
with open(save_dir+'/'+seq_name+'.txt','w') as f:
for img_l in sorted(img_dict.items(),key=lambda x:x[0]):
if args.dataset_name=='realBR' or args.dataset_name=='GOPRO-VFI_copy':
gt_dict_psnr = sorted(img_l[1].items(),key=lambda x:x[0])
gt_dict_ssim = sorted(ssim_dict[seq_name][img_l[0]].items(),key=lambda x:x[0])
gt_dict_lpips = sorted(lpips_dict[seq_name][img_l[0]].items(),key=lambda x:x[0])
for xx, yy, zz in zip(gt_dict_psnr, gt_dict_ssim,gt_dict_lpips):
assert xx[0] == yy[0] == zz[0]
f.write( xx[0]+'\t'+ xx[1]+'\t'+ yy[1]+'\t'+zz[1]+'\n')
elif args.dataset_name == 'GOPRO-Dual':
img_name = img_l[0]
for gt_l in sorted(img_l[1].items(),key=lambda x:x[0]):
img_psnr_record = gt_l[1]
img_ssim_record = ssim_dict[seq_name][img_name][gt_l[0]]
img_lpips_record = lpips_dict[seq_name][img_name][gt_l[0]]
f.write(img_name+'\t'+gt_l[0]+'\t'+img_psnr_record+'\t'+img_ssim_record+'\t'+img_lpips_record+'\n')
else:
raise Exception('Not supported data!!!')
with open(save_dir+'/overall_metrics.txt','w') as f:
f.write('Overall PSNR: %.4f\n'%(np.mean(psnr_list)))
f.write('Overall SSIM: %.4f\n'%(np.mean(ssim_list)))
f.write('Overall LPIPS: %.4f\n'%(np.mean(lpips_list)))
f.write('metrics by time stamp:\n')
psnr_time = sorted(psnr_time.items(),key=lambda x:float(x[0]))
for kk in psnr_time:
avg_psnr = format(np.mean(kk[1]),'.4f')
avg_ssim = format(np.mean(ssim_time[kk[0]]),'.4f')
avg_lpips = format(np.mean(lpips_time[kk[0]]),'.4f')
f.write( 'tiemstamp:'+str(kk[0])+'\t'+ avg_psnr+'\t'+ avg_ssim+'\t'+avg_lpips+'\n')
print('---------------------------------------------------------------')
print('Overall PSNR: %.4f'%(np.mean(psnr_list)))
print('Overall SSIM: %.4f'%(np.mean(ssim_list)))
print('Overall LPIPS: %.4f'%(np.mean(lpips_list)))
print('---------------------------------------------------------------')
if __name__ == "__main__":
# For reproduction
seed = 1234
random.seed(seed)
np.random.seed(seed)
torch.manual_seed(seed)
torch.cuda.manual_seed_all(seed)
model = Model(config=args)
test(model)