-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathconverted_det.py
166 lines (133 loc) · 6.44 KB
/
converted_det.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
import torch
import torch.nn as nn
import matplotlib
matplotlib.use('agg')
import warnings
warnings.filterwarnings('ignore')
import numpy as np
from tqdm import tqdm
from copy import deepcopy
import matplotlib.pyplot as plt
import argparse, logging, datetime, time, os
from src.utils import seed_all, accuracy, AverageMeter, result2csv, setup_default_logging, mergeConvBN
from src.clipquantization import replace_relu_by_cqrelu
from src.convertor import *
from src.dataset import *
from src.model import *
from src.det.data.voc import VOCDetection
from src.det.data.coco import COCODataset
from src.det.data.transforms import TrainTransforms, ColorTransforms, ValTransforms
from src.det.evaluator.cocoapi_evaluator import COCOAPIEvaluator
from src.det.evaluator.vocapi_evaluator import VOCAPIEvaluator
from src.det.utils.com_flops_params import FLOPs_and_Params
parser = argparse.ArgumentParser(description='Conversion')
parser.add_argument('--save_result', type=bool, default=True)
parser.add_argument('--device', type=str, default='3')
parser.add_argument('--use_cuda', type=bool, default=True)
parser.add_argument('--dataset', type=str, default='voc', choices=['voc', 'coco'])
parser.add_argument('--model', type=str, default='VGG16', choices=['ResNet50', 'VGG16'])
'''
/data/ly/casc/20221102-100752-yolo-VGG16-voc-True-4-42
/data/ly/casc/20221102-100832-yolo-VGG16-voc-True-8-42/
/data/ly/casc/20221102-100917-yolo-VGG16-voc-True-16-42
/data/ly/casc/20221102-101016-yolo-VGG16-voc-True-32-42
/data/ly/casc/20221031-234132-yolo-ResNet50-voc-True-4-42
/data/ly/casc/20221031-092006-yolo-ResNet50-voc-True-8-42
/data/ly/casc/20221030-110639-yolo-ResNet50-voc-True-16-42
/data0/ly/casc/20221029-181623-yolo-ResNet50-voc-True-32-42
/data/ly/casc/20221114-135924-yolo-VGG16-coco-True-4-42
/data/ly/casc/20221114-135924-yolo-VGG16-coco-True-8-42
/data/ly/casc/20221114-135924-yolo-VGG16-coco-True-16-42
'''
parser.add_argument('--ckpt_path', type=str, default="/data/ly/casc/20221114-135924-yolo-VGG16-coco-True-4-42")
parser.add_argument('--neg', default=True, type=bool, help='negtive spikes')
# parser.add_argument('--neg', action='store_true', help='negtive spikes')
parser.add_argument('--sleep', default=4, type=int, help='sleep time')
parser.add_argument('--margin', default=4, type=float, help='sleep margin')
parser.add_argument('--T', default=4, type=int, help='simulation time')
parser.add_argument('--cqrelu', type=bool, default=True)
# parser.add_argument('--cqrelu', action='store_true')
parser.add_argument('--qlevel', type=int, default=4)
parser.add_argument('--data_path', type=str, default='/data/datasets', help='/Users/lee/data/datasets')
parser.add_argument('--saved_dir', type=str, default='/data/ly/casc/conversion/')
parser.add_argument('--saved_csv', type=str, default='./result_det_conversion.csv')
parser.add_argument('--train_batch', default=30, type=int, help='batch size for get max')
parser.add_argument('--batch_size', default=8, type=int, help='batch size for testing')
parser.add_argument('--img_size', type=int, default=640, help='The size of input image')
parser.add_argument('--seed', default=14, type=int, help='seed')
parser.add_argument('--p', default=0.999, type=float, help='percentile for data normalization. 0-1')
parser.add_argument('--gamma', default=1, type=int, help='burst spike and max spikes IF can emit')
parser.add_argument('--lipool', default=True, type=bool, help='LIPooling')
parser.add_argument('--soft_mode', default=True, type=bool, help='soft_reset')
parser.add_argument('--channelnorm', default=False, type=bool, help='channelnorm')
parser.add_argument('--pseudo_convert', default=False, type=bool, help='pseudo_convert')
parser.add_argument('--merge', default=True, type=bool, help='fuseConvBN')
args = parser.parse_args()
if __name__ == '__main__':
args.backbone = args.ckpt_path.split('-')[-5]
args.dataset = args.ckpt_path.split('-')[-4]
args.cqrelu = args.ckpt_path.split('-')[-3] == 'True'
args.qlevel = int(args.ckpt_path.split('-')[-2])
seed_all(seed=args.seed)
device = torch.device("cuda:%s" % args.device) if args.use_cuda else 'cpu'
train_size = val_size = args.img_size
if args.dataset == 'voc':
# 加载voc数据集
data_dir = '/data/datasets/VOC_2007/VOCdevkit/'
num_classes = 20
evaluator = VOCAPIEvaluator(
data_dir=data_dir,
img_size=val_size,
device=device,
transform=ValTransforms(val_size))
elif args.dataset == 'coco':
# 加载COCO数据集
data_dir = '/data/datasets/COCO/'
num_classes = 80
evaluator = COCOAPIEvaluator(
data_dir=data_dir,
img_size=val_size,
device=device,
transform=ValTransforms(val_size)
)
else:
args.logger.info('unknow dataset !! Only support voc and coco !!')
exit(0)
from src.det.models.yolo import YOLO
yolo_net = YOLO(device, img_size=train_size, num_classes=num_classes, trainable=True, center_sample=False, backbone=args.backbone)
model = yolo_net
model = model.to(device)
if args.cqrelu:
model = replace_relu_by_cqrelu(model, args.qlevel)
model.trainable = False
model.eval()
FLOPs_and_Params(model=model,
img_size=train_size,
device=device)
model.load_state_dict(torch.load(os.path.join(args.ckpt_path, 'checkpoints.pth'), map_location=device))
model = mergeConvBN(model) if args.merge else model
model_eval = model
model_eval.trainable = False
model_eval.set_grid(val_size)
model_eval.eval()
evaluator.evaluate(model_eval, args.batch_size)
args.map = evaluator.map
print("map: %.6f"% args.map)
if args.cqrelu:
converter = CQConvertor(
soft_mode=args.soft_mode,
lipool=args.lipool,
gamma=args.gamma,
pseudo_convert=args.pseudo_convert,
merge=args.merge,
neg=args.neg,
sleep_time=[args.qlevel, args.qlevel+args.sleep])
snn = converter(deepcopy(model))
maps = evaluator.evaluate_T(snn, args.batch_size, args.T, args)
args_dict = vars(args)
args_dict.update({"map_best": max(maps)})
args_dict.update({"best_ind": np.argmax(maps)})
for i in range(len(maps)):
args_dict.update({"acc_%d" % i: maps[i]})
if args.save_result:
result2csv(args.saved_csv, args_dict)