-
Notifications
You must be signed in to change notification settings - Fork 288
/
demo.py
204 lines (170 loc) · 8.48 KB
/
demo.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
import argparse
import ast
import pprint
import mxnet as mx
from mxnet.module import Module
from symdata.bbox import im_detect
from symdata.loader import load_test, generate_batch
from symdata.vis import vis_detection
from symnet.model import load_param, check_shape
def demo_net(sym, class_names, args):
# print config
print('called with args\n{}'.format(pprint.pformat(vars(args))))
# setup context
if args.gpu:
ctx = mx.gpu(int(args.gpu))
else:
ctx = mx.cpu(0)
# load single test
im_tensor, im_info, im_orig = load_test(args.image, short=args.img_short_side, max_size=args.img_long_side,
mean=args.img_pixel_means, std=args.img_pixel_stds)
# generate data batch
data_batch = generate_batch(im_tensor, im_info)
# load params
arg_params, aux_params = load_param(args.params, ctx=ctx)
# produce shape max possible
data_names = ['data', 'im_info']
label_names = None
data_shapes = [('data', (1, 3, args.img_long_side, args.img_long_side)), ('im_info', (1, 3))]
label_shapes = None
# check shapes
check_shape(sym, data_shapes, arg_params, aux_params)
# create and bind module
mod = Module(sym, data_names, label_names, context=ctx)
mod.bind(data_shapes, label_shapes, for_training=False)
mod.init_params(arg_params=arg_params, aux_params=aux_params)
# forward
mod.forward(data_batch)
rois, scores, bbox_deltas = mod.get_outputs()
rois = rois[:, 1:]
scores = scores[0]
bbox_deltas = bbox_deltas[0]
im_info = im_info[0]
# decode detection
det = im_detect(rois, scores, bbox_deltas, im_info,
bbox_stds=args.rcnn_bbox_stds, nms_thresh=args.rcnn_nms_thresh,
conf_thresh=args.rcnn_conf_thresh)
# print out
for [cls, conf, x1, y1, x2, y2] in det:
if cls > 0 and conf > args.vis_thresh:
print(class_names[int(cls)], conf, [x1, y1, x2, y2])
# if vis
if args.vis:
vis_detection(im_orig, det, class_names, thresh=args.vis_thresh)
def parse_args():
parser = argparse.ArgumentParser(description='Demonstrate a Faster R-CNN network',
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument('--network', type=str, default='vgg16', help='base network')
parser.add_argument('--params', type=str, default='', help='path to trained model')
parser.add_argument('--dataset', type=str, default='voc', help='training dataset')
parser.add_argument('--image', type=str, default='', help='path to test image')
parser.add_argument('--gpu', type=str, default='', help='gpu device eg. 0')
parser.add_argument('--vis', action='store_true', help='display results')
parser.add_argument('--vis-thresh', type=float, default=0.7, help='threshold display boxes')
# faster rcnn params
parser.add_argument('--img-short-side', type=int, default=600)
parser.add_argument('--img-long-side', type=int, default=1000)
parser.add_argument('--img-pixel-means', type=str, default='(0.0, 0.0, 0.0)')
parser.add_argument('--img-pixel-stds', type=str, default='(1.0, 1.0, 1.0)')
parser.add_argument('--rpn-feat-stride', type=int, default=16)
parser.add_argument('--rpn-anchor-scales', type=str, default='(8, 16, 32)')
parser.add_argument('--rpn-anchor-ratios', type=str, default='(0.5, 1, 2)')
parser.add_argument('--rpn-pre-nms-topk', type=int, default=6000)
parser.add_argument('--rpn-post-nms-topk', type=int, default=300)
parser.add_argument('--rpn-nms-thresh', type=float, default=0.7)
parser.add_argument('--rpn-min-size', type=int, default=16)
parser.add_argument('--rcnn-num-classes', type=int, default=21)
parser.add_argument('--rcnn-feat-stride', type=int, default=16)
parser.add_argument('--rcnn-pooled-size', type=str, default='(14, 14)')
parser.add_argument('--rcnn-batch-size', type=int, default=1)
parser.add_argument('--rcnn-bbox-stds', type=str, default='(0.1, 0.1, 0.2, 0.2)')
parser.add_argument('--rcnn-nms-thresh', type=float, default=0.3)
parser.add_argument('--rcnn-conf-thresh', type=float, default=1e-3)
args = parser.parse_args()
args.img_pixel_means = ast.literal_eval(args.img_pixel_means)
args.img_pixel_stds = ast.literal_eval(args.img_pixel_stds)
args.rpn_anchor_scales = ast.literal_eval(args.rpn_anchor_scales)
args.rpn_anchor_ratios = ast.literal_eval(args.rpn_anchor_ratios)
args.rcnn_pooled_size = ast.literal_eval(args.rcnn_pooled_size)
args.rcnn_bbox_stds = ast.literal_eval(args.rcnn_bbox_stds)
return args
def get_voc_names(args):
from symimdb.pascal_voc import PascalVOC
args.rcnn_num_classes = len(PascalVOC.classes)
return PascalVOC.classes
def get_coco_names(args):
from symimdb.coco import coco
args.rcnn_num_classes = len(coco.classes)
return coco.classes
def get_vgg16_test(args):
from symnet.symbol_vgg import get_vgg_test
if not args.params:
args.params = 'model/vgg16-0010.params'
args.img_pixel_means = (123.68, 116.779, 103.939)
args.img_pixel_stds = (1.0, 1.0, 1.0)
args.net_fixed_params = ['conv1', 'conv2']
args.rpn_feat_stride = 16
args.rcnn_feat_stride = 16
args.rcnn_pooled_size = (7, 7)
return get_vgg_test(anchor_scales=args.rpn_anchor_scales, anchor_ratios=args.rpn_anchor_ratios,
rpn_feature_stride=args.rpn_feat_stride, rpn_pre_topk=args.rpn_pre_nms_topk,
rpn_post_topk=args.rpn_post_nms_topk, rpn_nms_thresh=args.rpn_nms_thresh,
rpn_min_size=args.rpn_min_size,
num_classes=args.rcnn_num_classes, rcnn_feature_stride=args.rcnn_feat_stride,
rcnn_pooled_size=args.rcnn_pooled_size, rcnn_batch_size=args.rcnn_batch_size)
def get_resnet50_test(args):
from symnet.symbol_resnet import get_resnet_test
if not args.params:
args.params = 'model/resnet50-0010.params'
args.img_pixel_means = (0.0, 0.0, 0.0)
args.img_pixel_stds = (1.0, 1.0, 1.0)
args.rpn_feat_stride = 16
args.rcnn_feat_stride = 16
args.rcnn_pooled_size = (14, 14)
return get_resnet_test(anchor_scales=args.rpn_anchor_scales, anchor_ratios=args.rpn_anchor_ratios,
rpn_feature_stride=args.rpn_feat_stride, rpn_pre_topk=args.rpn_pre_nms_topk,
rpn_post_topk=args.rpn_post_nms_topk, rpn_nms_thresh=args.rpn_nms_thresh,
rpn_min_size=args.rpn_min_size,
num_classes=args.rcnn_num_classes, rcnn_feature_stride=args.rcnn_feat_stride,
rcnn_pooled_size=args.rcnn_pooled_size, rcnn_batch_size=args.rcnn_batch_size,
units=(3, 4, 6, 3), filter_list=(256, 512, 1024, 2048))
def get_resnet101_test(args):
from symnet.symbol_resnet import get_resnet_test
if not args.params:
args.params = 'model/resnet101-0010.params'
args.img_pixel_means = (0.0, 0.0, 0.0)
args.img_pixel_stds = (1.0, 1.0, 1.0)
args.rpn_feat_stride = 16
args.rcnn_feat_stride = 16
args.rcnn_pooled_size = (14, 14)
return get_resnet_test(anchor_scales=args.rpn_anchor_scales, anchor_ratios=args.rpn_anchor_ratios,
rpn_feature_stride=args.rpn_feat_stride, rpn_pre_topk=args.rpn_pre_nms_topk,
rpn_post_topk=args.rpn_post_nms_topk, rpn_nms_thresh=args.rpn_nms_thresh,
rpn_min_size=args.rpn_min_size,
num_classes=args.rcnn_num_classes, rcnn_feature_stride=args.rcnn_feat_stride,
rcnn_pooled_size=args.rcnn_pooled_size, rcnn_batch_size=args.rcnn_batch_size,
units=(3, 4, 23, 3), filter_list=(256, 512, 1024, 2048))
def get_class_names(dataset, args):
datasets = {
'voc': get_voc_names,
'coco': get_coco_names
}
if dataset not in datasets:
raise ValueError("dataset {} not supported".format(dataset))
return datasets[dataset](args)
def get_network(network, args):
networks = {
'vgg16': get_vgg16_test,
'resnet50': get_resnet50_test,
'resnet101': get_resnet101_test
}
if network not in networks:
raise ValueError("network {} not supported".format(network))
return networks[network](args)
def main():
args = parse_args()
class_names = get_class_names(args.dataset, args)
sym = get_network(args.network, args)
demo_net(sym, class_names, args)
if __name__ == '__main__':
main()