blob: f8f202153322183ae8ce009f6157eecc72165304 [file] [log] [blame]
John Richardson8de92612018-02-22 14:09:31 +00001/*
2 * Copyright (c) 2018 ARM Limited.
3 *
4 * SPDX-License-Identifier: MIT
5 *
6 * Permission is hereby granted, free of charge, to any person obtaining a copy
7 * of this software and associated documentation files (the "Software"), to
8 * deal in the Software without restriction, including without limitation the
9 * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
10 * sell copies of the Software, and to permit persons to whom the Software is
11 * furnished to do so, subject to the following conditions:
12 *
13 * The above copyright notice and this permission notice shall be included in all
14 * copies or substantial portions of the Software.
15 *
16 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
17 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
18 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
19 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
20 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
21 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
22 * SOFTWARE.
23 */
24#ifndef ARM_COMPUTE_TEST_OPTICAL_FLOW
25#define ARM_COMPUTE_TEST_OPTICAL_FLOW
26
27#include "arm_compute/core/PyramidInfo.h"
28#include "arm_compute/core/TensorShape.h"
29#include "arm_compute/core/Types.h"
30#include "tests/AssetsLibrary.h"
31#include "tests/Globals.h"
32#include "tests/IAccessor.h"
33#include "tests/Types.h"
34#include "tests/framework/Asserts.h"
35#include "tests/framework/Fixture.h"
36#include "tests/validation/reference/OpticalFlow.h"
37
38namespace arm_compute
39{
40namespace test
41{
42namespace validation
43{
44template <typename TensorType,
45 typename AccessorType,
46 typename ArrayType,
47 typename ArrayAccessorType,
48 typename FunctionType,
49 typename PyramidType,
50 typename PyramidFunctionType,
51 typename T>
52
53class OpticalFlowValidationFixture : public framework::Fixture
54{
55public:
56 template <typename...>
57 void setup(std::string old_image_name, std::string new_image_name, OpticalFlowParameters params,
58 size_t num_levels, size_t num_keypoints, Format format, BorderMode border_mode)
59 {
60 std::mt19937 gen(library->seed());
61 std::uniform_int_distribution<uint8_t> int_dist(0, 255);
62 const uint8_t constant_border_value = int_dist(gen);
63
64 // Create keypoints
65 std::vector<KeyPoint> old_keypoints = generate_random_keypoints(library->get_image_shape(old_image_name), num_keypoints, library->seed(), num_levels);
66 std::vector<KeyPoint> new_keypoints_estimates = old_keypoints;
67
68 _target = compute_target(old_image_name, new_image_name, params, num_levels, old_keypoints, new_keypoints_estimates, format, border_mode, constant_border_value);
69 _reference = compute_reference(old_image_name, new_image_name, params, num_levels, old_keypoints, new_keypoints_estimates, format, border_mode, constant_border_value);
70 }
71
72protected:
73 template <typename V>
74 void fill(V &&tensor, const std::string image, Format format)
75 {
76 library->fill(tensor, image, format);
77 }
78
79 ArrayType compute_target(std::string old_image_name, std::string new_image_name, OpticalFlowParameters params, size_t num_levels,
80 std::vector<KeyPoint> &old_keypoints, std::vector<KeyPoint> &new_keypoints_estimates,
81 Format format, BorderMode border_mode, uint8_t constant_border_value)
82 {
83 // Get image shapes
84 TensorShape old_shape = library->get_image_shape(old_image_name);
85 TensorShape new_shape = library->get_image_shape(new_image_name);
86
87 // Create tensors
88 auto old_image = create_tensor<TensorType>(old_shape, format);
89 auto new_image = create_tensor<TensorType>(new_shape, format);
90
91 // Load keypoints
92 ArrayType old_points(old_keypoints.size());
93 ArrayType new_points_estimates(new_keypoints_estimates.size());
94 ArrayType new_points(old_keypoints.size());
95
96 fill_array(ArrayAccessorType(old_points), old_keypoints);
97 fill_array(ArrayAccessorType(new_points_estimates), new_keypoints_estimates);
98
99 // Create pyramid images
100 PyramidInfo pyramid_info(num_levels, SCALE_PYRAMID_HALF, old_image.info()->tensor_shape(), format);
101 PyramidType old_pyramid = create_pyramid<PyramidType>(pyramid_info);
102 PyramidType new_pyramid = create_pyramid<PyramidType>(pyramid_info);
103
104 // Create and configure pyramid functions
105 PyramidFunctionType old_gp;
106 old_gp.configure(&old_image, &old_pyramid, border_mode, constant_border_value);
107
108 PyramidFunctionType new_gp;
109 new_gp.configure(&new_image, &new_pyramid, border_mode, constant_border_value);
110
111 for(size_t i = 0; i < pyramid_info.num_levels(); ++i)
112 {
113 ARM_COMPUTE_EXPECT(old_pyramid.get_pyramid_level(i)->info()->is_resizable(), framework::LogLevel::ERRORS);
114 ARM_COMPUTE_EXPECT(new_pyramid.get_pyramid_level(i)->info()->is_resizable(), framework::LogLevel::ERRORS);
115 }
116
117 // Create and configure optical flow function
118 FunctionType optical_flow;
119
120 optical_flow.configure(&old_pyramid,
121 &new_pyramid,
122 &old_points,
123 &new_points_estimates,
124 &new_points,
125 params.termination,
126 params.epsilon,
127 params.num_iterations,
128 params.window_dimension,
129 params.use_initial_estimate,
130 border_mode,
131 constant_border_value);
132
133 ARM_COMPUTE_EXPECT(old_image.info()->is_resizable(), framework::LogLevel::ERRORS);
134 ARM_COMPUTE_EXPECT(new_image.info()->is_resizable(), framework::LogLevel::ERRORS);
135
136 // Allocate input tensors
137 old_image.allocator()->allocate();
138 new_image.allocator()->allocate();
139
140 // Allocate pyramids
141 old_pyramid.allocate();
142 new_pyramid.allocate();
143
144 ARM_COMPUTE_EXPECT(!old_image.info()->is_resizable(), framework::LogLevel::ERRORS);
145 ARM_COMPUTE_EXPECT(!new_image.info()->is_resizable(), framework::LogLevel::ERRORS);
146
147 for(size_t i = 0; i < pyramid_info.num_levels(); ++i)
148 {
149 ARM_COMPUTE_EXPECT(!old_pyramid.get_pyramid_level(i)->info()->is_resizable(), framework::LogLevel::ERRORS);
150 ARM_COMPUTE_EXPECT(!new_pyramid.get_pyramid_level(i)->info()->is_resizable(), framework::LogLevel::ERRORS);
151 }
152
153 // Fill tensors
154 fill(AccessorType(old_image), old_image_name, format);
155 fill(AccessorType(new_image), new_image_name, format);
156
157 // Compute functions
158 old_gp.run();
159 new_gp.run();
160 optical_flow.run();
161
162 return new_points;
163 }
164
165 std::vector<KeyPoint> compute_reference(std::string old_image_name, std::string new_image_name,
166 OpticalFlowParameters params, size_t num_levels,
167 std::vector<KeyPoint> &old_keypoints, std::vector<KeyPoint> &new_keypoints_estimates,
168 Format format, BorderMode border_mode, uint8_t constant_border_value)
169 {
170 SimpleTensor<T> old_image{ library->get_image_shape(old_image_name), data_type_from_format(format) };
171 SimpleTensor<T> new_image{ library->get_image_shape(new_image_name), data_type_from_format(format) };
172
173 fill(old_image, old_image_name, format);
174 fill(new_image, new_image_name, format);
175
176 return reference::optical_flow<T>(old_image, new_image, params, num_levels, old_keypoints, new_keypoints_estimates,
177 border_mode, constant_border_value);
178 }
179
180 ArrayType _target{};
181 std::vector<KeyPoint> _reference{};
182};
183} // namespace validation
184} // namespace test
185} // namespace arm_compute
186#endif /* ARM_COMPUTE_TEST_OPTICAL_FLOW */