Commit
·
a25806a
1
Parent(s):
b89c008
Delete test_multicard.py
Browse files- test_multicard.py +0 -99
test_multicard.py
DELETED
@@ -1,99 +0,0 @@
|
|
1 |
-
import time, os, torch, argparse, warnings, glob, pandas, json
|
2 |
-
|
3 |
-
from utils.tools import *
|
4 |
-
from dlhammer import bootstrap
|
5 |
-
|
6 |
-
from dataLoader_multiperson import val_loader
|
7 |
-
from loconet import loconet
|
8 |
-
|
9 |
-
|
10 |
-
class DataPrep():
|
11 |
-
|
12 |
-
def __init__(self, cfg):
|
13 |
-
self.cfg = cfg
|
14 |
-
|
15 |
-
def val_dataloader(self):
|
16 |
-
cfg = self.cfg
|
17 |
-
loader = val_loader(cfg, trialFileName = cfg.evalTrialAVA, \
|
18 |
-
audioPath = os.path.join(cfg.audioPathAVA , cfg.evalDataType), \
|
19 |
-
visualPath = os.path.join(cfg.visualPathAVA, cfg.evalDataType), \
|
20 |
-
num_speakers=cfg.MODEL.NUM_SPEAKERS,
|
21 |
-
)
|
22 |
-
valLoader = torch.utils.data.DataLoader(loader,
|
23 |
-
batch_size=cfg.VAL.BATCH_SIZE,
|
24 |
-
shuffle=False,
|
25 |
-
num_workers=16)
|
26 |
-
return valLoader
|
27 |
-
|
28 |
-
|
29 |
-
def prepare_context_files(cfg):
|
30 |
-
path = os.path.join(cfg.DATA.dataPathAVA, "csv")
|
31 |
-
for phase in ["val", "test"]:
|
32 |
-
csv_f = f"{phase}_loader.csv"
|
33 |
-
csv_orig = f"{phase}_orig.csv"
|
34 |
-
entity_f = os.path.join(path, phase + "_entity.json")
|
35 |
-
ts_f = os.path.join(path, phase + "_ts.json")
|
36 |
-
if os.path.exists(entity_f) and os.path.exists(ts_f):
|
37 |
-
continue
|
38 |
-
orig_df = pandas.read_csv(os.path.join(path, csv_orig))
|
39 |
-
entity_data = {}
|
40 |
-
ts_to_entity = {}
|
41 |
-
|
42 |
-
for index, row in orig_df.iterrows():
|
43 |
-
|
44 |
-
entity_id = row['entity_id']
|
45 |
-
video_id = row['video_id']
|
46 |
-
if row['label'] == "SPEAKING_AUDIBLE":
|
47 |
-
label = 1
|
48 |
-
else:
|
49 |
-
label = 0
|
50 |
-
ts = float(row['frame_timestamp'])
|
51 |
-
if video_id not in entity_data.keys():
|
52 |
-
entity_data[video_id] = {}
|
53 |
-
if entity_id not in entity_data[video_id].keys():
|
54 |
-
entity_data[video_id][entity_id] = {}
|
55 |
-
if ts not in entity_data[video_id][entity_id].keys():
|
56 |
-
entity_data[video_id][entity_id][ts] = []
|
57 |
-
|
58 |
-
entity_data[video_id][entity_id][ts] = label
|
59 |
-
|
60 |
-
if video_id not in ts_to_entity.keys():
|
61 |
-
ts_to_entity[video_id] = {}
|
62 |
-
if ts not in ts_to_entity[video_id].keys():
|
63 |
-
ts_to_entity[video_id][ts] = []
|
64 |
-
ts_to_entity[video_id][ts].append(entity_id)
|
65 |
-
|
66 |
-
with open(entity_f) as f:
|
67 |
-
json.dump(entity_data, f)
|
68 |
-
|
69 |
-
with open(ts_f) as f:
|
70 |
-
json.dump(ts_to_entity, f)
|
71 |
-
|
72 |
-
|
73 |
-
def main():
|
74 |
-
cfg = bootstrap(print_cfg=False)
|
75 |
-
print(cfg)
|
76 |
-
epoch = cfg.RESUME_EPOCH
|
77 |
-
|
78 |
-
warnings.filterwarnings("ignore")
|
79 |
-
|
80 |
-
cfg = init_args(cfg)
|
81 |
-
|
82 |
-
data = DataPrep(cfg)
|
83 |
-
|
84 |
-
prepare_context_files(cfg)
|
85 |
-
|
86 |
-
if cfg.downloadAVA == True:
|
87 |
-
preprocess_AVA(cfg)
|
88 |
-
quit()
|
89 |
-
|
90 |
-
s = loconet(cfg)
|
91 |
-
|
92 |
-
s.loadParameters(cfg.RESUME_PATH)
|
93 |
-
mAP = s.evaluate_network(epoch=epoch, loader=data.val_dataloader())
|
94 |
-
print(f"evaluate ckpt: {cfg.RESUME_PATH}")
|
95 |
-
print(mAP)
|
96 |
-
|
97 |
-
|
98 |
-
if __name__ == '__main__':
|
99 |
-
main()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|