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
|
# Copyright 2020 The Chromium Authors
# Use of this source code is governed by a BSD-style license that can be
# found in the LICENSE file.
import logging
import gold_inexact_matching.iterative_parameter_optimizer\
as iterative_optimizer
from gold_inexact_matching import parameter_set
class BruteForceParameterOptimizer(
iterative_optimizer.IterativeParameterOptimizer):
"""A ParameterOptimizer for use with any number of changing parameters.
VERY slow, but provides the most complete information when trying to
optimize multiple parameters.
"""
def _RunOptimizationImpl(self) -> None:
# Look for the minimum max_delta that results in a successful comparison
# for each possible edge_threshold/max_diff combination.
#
# range/xrange(x, y) returns the range in [x, y), so adjust by 1 to make the
# range inclusive.
# We go from max to min instead of min to max for edge_threshold, so
# decrease the min by 1 instead of increasing the max by 1.
for edge_threshold in range(self._args.max_edge_threshold,
self._args.min_edge_threshold - 1,
-1 * self._args.edge_threshold_step):
should_continue = True
for max_diff in range(self._args.min_max_diff,
self._args.max_max_diff + 1,
self._args.max_diff_step):
for max_delta in range(self._args.min_delta_threshold,
self._args.max_delta_threshold + 1,
self._args.delta_threshold_step):
parameters = parameter_set.ParameterSet(max_diff, max_delta,
edge_threshold)
success, _, _ = self._RunComparisonForParameters(parameters)
if success:
print(f'Found good parameters {parameters}')
should_continue = False
break
logging.info('Found bad parameters %s', parameters)
# Increasing the max_diff for a given edge_threshold once we've found
# a good max_delta won't give us any new information, so go on to the
# next edge_threshold.
if not should_continue:
break
|
