/
vop_eval.py
167 lines (131 loc) · 6.24 KB
/
vop_eval.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
import numpy as np
from glob import glob
import json
from tqdm import tqdm
from PIL import Image
import os.path as osp
from utils import *
import argparse
def eval_iou(res_list,seq_list,ref_list,obj_class_dict,eval_decay=False):
# mask iou
thing_seen_miou_list = []
stuff_seen_miou_list = []
thing_unseen_miou_list = []
stuff_unseen_miou_list = []
# boundary iou
thing_seen_biou_list = []
stuff_seen_biou_list = []
thing_unseen_biou_list = []
stuff_unseen_biou_list = []
# decay
if eval_decay:
iou_decay_dict = {}
for i in range(80):
iou_decay_dict[i] = []
vp = VIPOSeg()
for s,r,f in tqdm(zip(seq_list,res_list,ref_list)):
video_id = s.split('/')[-1]
label_list = sorted(glob(s+'/*'))
pred_list = sorted(glob(r+'/*'))
ann_list = sorted(glob(f+'/*'))
ann_name_list = [x.split('/')[-1] for x in ann_list]
assert len(label_list) == len(pred_list), 'incomplete label/pred'
obj_ids = []
for i in range(len(label_list)):
label = Image.open(label_list[i])
pred = Image.open(pred_list[i])
label = np.array(label,np.uint8)
pred = np.array(pred,np.uint8)
obj_num = len(obj_ids)
for id in obj_ids:
mask_gt = label==id
mask_pred = pred==id
# mask iou and boundary iou
if (np.sum(mask_pred) == 0) and (np.sum(mask_gt) != 0):
miou = 0.
biou = 0.
elif (np.sum(mask_pred) != 0) and (np.sum(mask_gt) == 0):
miou = 0.
biou = 0.
elif (np.sum(mask_pred) ==0) and (np.sum(mask_gt) ==0):
miou = 1.
biou = 1.
else:
miou = np.sum(mask_gt & mask_pred) / np.sum(mask_gt | mask_pred)
biou = boundary_iou(mask_gt.astype(np.uint8),mask_pred.astype(np.uint8),dilation_ratio=0.02)
class_id = int(obj_class_dict[video_id][str(id)])
if class_id == 98:
if video_id in vp.other_machine_videos:
stuff_unseen_miou_list.append(miou)
stuff_unseen_biou_list.append(biou)
else:
stuff_seen_miou_list.append(miou)
stuff_seen_biou_list.append(biou)
elif class_id in vp.thing_unseen_class:
thing_unseen_miou_list.append(miou)
thing_unseen_biou_list.append(biou)
elif class_id in vp.stuff_unseen_class:
stuff_unseen_miou_list.append(miou)
stuff_unseen_biou_list.append(biou)
elif class_id in vp.thing_seen_class:
thing_seen_miou_list.append(miou)
thing_seen_biou_list.append(biou)
elif class_id in vp.stuff_seen_class:
stuff_seen_miou_list.append(miou)
stuff_seen_biou_list.append(biou)
if eval_decay:
iou_decay_dict[obj_num].append((miou+biou)/2.)
# exclude obj in ref frames, eval in next frame
frame_name = label_list[i].split('/')[-1]
if frame_name in ann_name_list:
ann_idx = ann_name_list.index(frame_name)
ann = Image.open(ann_list[ann_idx])
obj_ids.extend([x for x in np.unique(ann) if x!=0])
res_dict = {}
res_dict['thing_seen_miou'] = np.mean(thing_seen_miou_list)
res_dict['thing_unseen_miou'] = np.mean(thing_unseen_miou_list)
res_dict['stuff_seen_miou'] = np.mean(stuff_seen_miou_list)
res_dict['stuff_unseen_miou'] = np.mean(stuff_unseen_miou_list)
res_dict['thing_seen_biou'] = np.mean(thing_seen_biou_list)
res_dict['thing_unseen_biou'] = np.mean(thing_unseen_biou_list)
res_dict['stuff_seen_biou'] = np.mean(stuff_seen_biou_list)
res_dict['stuff_unseen_biou'] = np.mean(stuff_unseen_biou_list)
res_dict['thing_seen_iou'] = (res_dict['thing_seen_miou']+res_dict['thing_seen_biou'])/2
res_dict['thing_unseen_iou'] = (res_dict['thing_unseen_miou']+res_dict['thing_unseen_biou'])/2
res_dict['stuff_seen_iou'] = (res_dict['stuff_seen_miou']+res_dict['stuff_seen_biou'])/2
res_dict['stuff_unseen_iou'] = (res_dict['stuff_unseen_miou']+res_dict['stuff_unseen_biou'])/2
res_dict['overall_iou'] = (res_dict['thing_seen_iou']+res_dict['thing_unseen_iou']\
+res_dict['stuff_seen_iou']+res_dict['stuff_unseen_iou'])/4
if eval_decay:
x = []
y = []
for k in iou_decay_dict.keys():
v = iou_decay_dict[k]
if v!=[] and k<60:
x.append(k)
y.append(np.mean(v))
_x = np.expand_dims(np.array(x),-1)
_y = np.expand_dims(np.array(y),-1)
A = _x/100
b = -np.log(_y)
decay = np.dot(np.dot(np.linalg.inv(np.dot(A.T,A)),A.T),b)
res_dict['decay'] = decay[0,0]
return res_dict
if __name__=='__main__':
parser = argparse.ArgumentParser()
parser.add_argument('--data_path',type=str,default='./VIPOSeg/valid')
parser.add_argument('--res_path',type=str,required=True)
parser.add_argument('--eval_decay',action='store_true')
args = parser.parse_args()
res_list = sorted(glob(args.res_path+'/*'))
seq_list = sorted(glob(osp.join(args.data_path,'Annotations_gt')+'/*'))
ref_list = sorted(glob(osp.join(args.data_path,'Annotations')+'/*'))
assert len(res_list) > 0 and len(res_list) == len(res_list) and\
len(res_list) == len(ref_list), "{} results and {} data".format(len(res_list),len(ref_list))
# read obj_class.json
with open(osp.join(args.data_path,'obj_class.json'),'r') as f:
obj_class_dict = json.load(f)
res_dict = eval_iou(res_list,seq_list,ref_list,obj_class_dict,args.eval_decay)
for k in res_dict.keys():
v = res_dict[k]*100 if 'iou' in k else res_dict[k]
print("{}: {:.2f}".format(k,v))