/
different_scales.py
executable file
·83 lines (71 loc) · 2.89 KB
/
different_scales.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
#!/usr/bin/env python
# --------------------------------------------------------
# Fast R-CNN
# Copyright (c) 2015 Microsoft
# Licensed under The MIT License [see LICENSE for details]
# Written by Dennis Dam
# --------------------------------------------------------
"""Set up paths for Fast R-CNN."""
from os import chmod
from os.path import join, dirname, abspath
import sys
this_dir = dirname(abspath(__file__))
sys.path.insert(0,join(this_dir,'..','src'))
import _init_paths
from _init_paths import faster_rcnn_root, caffe_root
from scenario import Scenario
from network import RegionProposalNetworkConfig as RpnConfig, FastRcnnNetworkConfig as FastRcnnConfig
from solver import SolverConfig
from copy import deepcopy
import yaml
scenarios_dir = join(this_dir, 'scenarios')
def create_scenarios():
alt_opt_cfg = join(faster_rcnn_root, "experiments/cfgs/faster_rcnn_alt_opt.yml")
base_scenario = Scenario(
scenarios_dir=scenarios_dir,
scenario="scales_2_4_8",
train_imdb="technicaldrawings_single-numbers_train",
test_imdb="technicaldrawings_single-numbers_val",
weights_path=join(faster_rcnn_root, "data/imagenet_models/ZF.v2.caffemodel"), # you have to download this first
gpu_id=0,
max_iters=[1, 1, 1, 1], # max iters
rpn_config=RpnConfig(num_classes=2, anchor_scales=[8, 16, 32], anchor_feat_stride=16),
fast_rcnn_config=FastRcnnConfig(num_classes=2),
solver_config=SolverConfig(),
config=yaml.load(open(alt_opt_cfg))
)
small_scales = deepcopy(base_scenario)
small_scales.name("scales_4_8_16")
small_scales.rpn_config.anchor_scales = [4, 8, 16]
small_scales.fast_rcnn_config.anchor_scales = small_scales.rpn_config.anchor_scales
default_scales = deepcopy(base_scenario)
default_scales.name("scales_8_16_32")
default_scales.rpn_config.anchor_scales = [8, 16, 32]
default_scales.fast_rcnn_config.anchor_scales = default_scales.rpn_config.anchor_scales
large_scales = deepcopy(base_scenario)
large_scales.name("scales_16_32_64")
large_scales.rpn_config.anchor_scales = [16, 32, 64]
large_scales.fast_rcnn_config.anchor_scales = large_scales.rpn_config.anchor_scales
return small_scales, default_scales, large_scales
def generate():
small, default, large=create_scenarios()
for scenario in [small,default,large]:
print "*"*160
print 'Generating scenario:',scenario.scenario
print "*"*160
print scenario
scenario.generate()
run_all_script_path = join(scenarios_dir, 'run_all.sh')
run_all_script = open(run_all_script_path, 'wb')
run_all_script.write("""
{small} || true
{default} || true
{large} || true
""".format(
small=small.script_path,
default=default.script_path,
large=large.script_path)
)
chmod(run_all_script_path, 0755)
if __name__ == '__main__':
generate()