-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathpredict_compute_delta.py
175 lines (160 loc) · 8.08 KB
/
predict_compute_delta.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
import os
import numpy as np
import pickle
import random
import pandas as pd
import time
import argparse
import json
import sys
def parse_args():
parser = argparse.ArgumentParser('W2VVPP predictor')
parser.add_argument('testCollection', type=str,
help='test collection')
parser.add_argument('sim_name', type=str,
help='sub-folder where computed similarities are saved')
parser.add_argument('--rootpath', type=str, default='/data1/wzy/',
help='path to datasets. (default: %s)'%'/data1/wzy/')
parser.add_argument("--config_name", default="no", type=str,
help='congig')
parser.add_argument("--model_path", default="no", type=str,
help='None')
parser.add_argument("--predict_result_file", default="no", type=str,
help='predict_result_file')
parser.add_argument('--original_cap', type=str, default='testCollection.captionsubset_neginfo.txt',
help='original_cap')
parser.add_argument("--negated_cap", default="testCollection.caption.falseset_withneginfo.txt", type=str,
help='negated_cap')
args = parser.parse_args()
return args
def write_to_predict_result_file(
predict_result_file, config,model_path,
result_tuple, testCollection, name_str="Text to video"
):
"""
:param predict_result_file:
:param model_path:
:param checkpoint:
:param result_tuple: [(r1, r5, r10, medr, meanr, mir, mAP), ...]
:return:
"""
result_file_dir = os.path.dirname(predict_result_file)
if not os.path.exists(result_file_dir):
os.makedirs(result_file_dir)
print(predict_result_file)
with open(predict_result_file, 'a') as f:
(r1, r5, r10, medr, mir, mAP) = result_tuple
tempStr = " * %s:\n" % name_str
tempStr += " * dr_1_5_10: {}\n".format([round(r1, 3), round(r5, 3), round(r10, 3)])
tempStr += " * dmedr, dmir: {}\n".format([round(medr, 3), round(mir, 3)])
tempStr += " * dmAP: {}\n".format(round(mAP, 3))
tempStr += " * " + '-' * 10
print(tempStr)
f.write(str(time.asctime(time.localtime(time.time()))) + '\t')
f.write(model_path)
for each in [config, testCollection, round(r1, 3), round(r5, 3), round(r10, 3),
round(mir, 3)]:
f.write(str(each))
f.write('\t')
f.write('\n')
pass
def load_dict(predpath):
pkl_file = open(predpath, 'rb')
pred_data = pickle.load(pkl_file)
pred_data2={}
for k,v in pred_data.items():
kk="#".join(k.split("#")[:2])
pred_data2[kk]=v
#print((pred_data.keys()))
return pred_data
def compute_mean_delta(origin_metrics,false_cap_predpath):
pkl_file = open(false_cap_predpath, 'rb')
print(false_cap_predpath)
pred_data = pickle.load(pkl_file)
aps = np.zeros(len(pred_data))
r1s = np.zeros(len(pred_data))
r5s = np.zeros(len(pred_data))
r10s= np.zeros(len(pred_data))
ranks= np.zeros(len(pred_data))
for i, (idx, txt) in enumerate(pred_data.items()):
try:
capid = "#".join(idx.split("#")[:-1])
aps[i] = (origin_metrics[capid]["mAP"] - txt["mAP"])
except:
capid = "#".join(idx.split("#"))
aps[i] = (origin_metrics[capid]["mAP"] - txt["mAP"])
r1s[i]=origin_metrics[capid]["r1"]-txt["r1"]
r5s[i]=origin_metrics[capid]["r5"]-txt["r5"]
r10s[i]=origin_metrics[capid]["r10"]-txt["r10"]
ranks[i]=1/origin_metrics[capid]["ranks"]-1/txt["ranks"]
dmAP=aps.mean()
dr1 = r1s.mean()
dr5 = r5s.mean()
dr10=r10s.mean()
dmeanr=ranks.mean()
dmedr=np.median(ranks)
return dr1,dr5, dr10, dmedr, dmeanr, dmAP
def main():
opt = parse_args()
print(json.dumps(vars(opt), indent=2))
rootpath = opt.rootpath
model_path=opt.model_path
testCollection=opt.testCollection
model_name = os.path.basename(opt.model_path)
original_cap=opt.original_cap
negated_cap=opt.negated_cap
query_set = testCollection + ".caption.falseset.txt"
# subsetid= set(open(os.path.join(rootpath, testCollection,"TextData/negated_subset.txt")).read().strip().split("\n"))
false_cap_predpath = os.path.join(rootpath, testCollection, 'SimilarityIndex',
negated_cap, opt.sim_name,'t2v_eval.pkl')
# false_cap_predpath = os.path.join(rootpath, testCollection, 'SimilarityIndex', testCollection + ".caption.falseset.txt", config,'t2v_res.pkl')
# false_cap_predpath = os.path.join(rootpath, testCollection, 'SimilarityIndex',
# testCollection + ".caption.falseset_withneginfo.txt", config, params, 't2v_res.pkl')
# false_cap_predpath = "/home/wzy/VisualSearch/msrvtt10ktest/SimilarityIndex/msrvtt10ktest.caption.txt/clip4clip/msrvtt10ktest.falseset_msrvtt_retrieval_bs32_seqTransf_crEn.pkl"
origin_cap_predpath = os.path.join(rootpath, testCollection, 'SimilarityIndex',
original_cap, opt.sim_name,
't2v_eval.pkl')
# origin_cap_predpath = os.path.join(rootpath, testCollection, 'SimilarityIndex',testCollection + ".caption.txt", config,'t2v_res.pkl')
# origin_cap_predpath = os.path.join(rootpath, testCollection, 'SimilarityIndex',
# testCollection + ".captionsubset_neginfo.txt", config, params, 't2v_res.pkl')
result_file_dir = os.path.dirname(opt.predict_result_file)
result_file_name = os.path.basename(opt.predict_result_file)
# origin_cap_predpath="/home/wzy/VisualSearch/msrvtt10ktest/SimilarityIndex/msrvtt10ktest.caption.txt/clip4clip/msrvtt10ktest.msrvtt7k_retrieval_bs32_seqTransf_crEn.pkl"
origin_metrics = load_dict(origin_cap_predpath)
# dr1, dr5, dr10, dmedr, dmeanr, dmAP = compute_subset_delta(origin_metrics, false_cap_predpath,subsetid)
dr1, dr5, dr10, dmedr, dmeanr, dmAP = compute_mean_delta(origin_metrics, false_cap_predpath)
write_to_predict_result_file(
os.path.join(result_file_dir, 'TextToVideo', result_file_name), opt.config_name,opt.model_path,
(dr1, dr5, dr10, dmedr, dmeanr, dmAP), query_set
)
os.remove(origin_cap_predpath)
os.remove(false_cap_predpath)
# clip2video
#
# dirnames = ["msrvtt_data", "vatex_data"]
# testCollections = ['msrvtt1kAtest', "vatex_test1k5"]
# for dirname, testCollection in zip(dirnames, testCollections):
# query_set = testCollection + ".caption.falseset"
# # if testCollection=="vatex_test1k5":
# # params = "runs_7vatexreal_1_0.001_0.1_0.6_100_0.1_0.3_seed_2"
# # else:
# # params = "runs_7_1_0.001_0.1_0.6_100_0.1_0.3_seed_2"
# result_file_dir = os.path.join("/home/wzy/VisualSearch/CLIP2Video")
# false_cap_predpath = os.path.join(rootpath, "CLIP2Video", dirname, testCollection + ".caption.falseset_withneginfo",
# 't2v_res.pkl')
# origin_cap_predpath = os.path.join(rootpath, "CLIP2Video", dirname, testCollection + ".captionsubset_neginfo",
# 't2v_res.pkl')
# origin_metrics = load_dict(origin_cap_predpath)
# dr1, dr5, dr10, dmedr, dmeanr, dmAP = compute_mean_delta(origin_metrics, false_cap_predpath)
# write_to_predict_result_file(
# os.path.join(result_file_dir, "result_sotafalseset.txt"), "clip2video","",
# (dr1, dr5, dr10, dmedr, dmeanr, dmAP), query_set
# )
if __name__ == '__main__':
if len(sys.argv) == 1:
sys.argv = "predict_compute_delta.py msrvtt1kAtest CLIP.CLIPEnd2EndNegnomask/runs_10_1_0.001_0.1_0.6_100_0.1_0.3_seed_2 " \
"--model_path /data1/wzy/VisualSearch/msrvtt1kAtrain/w2vvpp_train/msrvtt1kAval/CLIP.CLIPEnd2EndNegnomask/runs_10_1_0.001_0.1_0.6_100_0.1_0.3_seed_2/checkpoint2.pth.tar " \
"--rootpath /data1/wzy/VisualSearch " \
"--predict_result_file result_log/result_test.txt " \
"--config_name CLIP.CLIPEnd2EndNegnomask".split(' ')
main()