blob: d2903fb849ef6aeadaf7c6481297461062fe3075 [file] [log] [blame]
Anthony Barbier6ff3b192017-09-04 18:44:23 +01001/*
2 * Copyright (c) 2016, 2017 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#include "arm_compute/runtime/CL/functions/CLFastCorners.h"
25
26#include "arm_compute/core/CL/OpenCL.h"
27#include "arm_compute/core/CL/kernels/CLFastCornersKernel.h"
28#include "arm_compute/core/Error.h"
29#include "arm_compute/core/TensorInfo.h"
30#include "arm_compute/core/Validate.h"
31#include "arm_compute/runtime/CL/CLScheduler.h"
32#include "arm_compute/runtime/ITensorAllocator.h"
33
34#include <algorithm>
35#include <cstring>
36
37using namespace arm_compute;
38
39CLFastCorners::CLFastCorners()
40 : _fast_corners_kernel(),
41 _suppr_func(),
42 _copy_array_kernel(),
43 _output(),
44 _suppr(),
45 _win(),
46 _non_max(false),
47 _num_corners(nullptr),
48 _num_buffer(),
49 _corners(nullptr),
50 _constant_border_value(0)
51{
52}
53
54void CLFastCorners::configure(const ICLImage *input, float threshold, bool nonmax_suppression, CLKeyPointArray *const corners,
55 unsigned int *num_corners, BorderMode border_mode, uint8_t constant_border_value)
56{
57 ARM_COMPUTE_ERROR_ON_TENSOR_NOT_2D(input);
58 ARM_COMPUTE_ERROR_ON(BorderMode::UNDEFINED != border_mode);
59 ARM_COMPUTE_ERROR_ON(nullptr == corners);
60 ARM_COMPUTE_ERROR_ON(threshold < 1 && threshold > 255);
61
62 TensorInfo tensor_info(input->info()->tensor_shape(), 1, DataType::U8);
63 _output.allocator()->init(tensor_info);
64
65 _non_max = nonmax_suppression;
66 _num_corners = num_corners;
67 _corners = corners;
68 _num_buffer = cl::Buffer(CLScheduler::get().context(), CL_MEM_ALLOC_HOST_PTR | CL_MEM_READ_WRITE, sizeof(unsigned int));
69 _constant_border_value = constant_border_value;
70
71 const bool update_number = (nullptr != _num_corners);
72
73 _fast_corners_kernel.configure(input, &_output, threshold, nonmax_suppression, border_mode);
74
75 if(!_non_max)
76 {
77 _copy_array_kernel.configure(&_output, update_number, corners, &_num_buffer);
78 }
79 else
80 {
81 _suppr.allocator()->init(tensor_info);
82
83 _suppr_func.configure(&_output, &_suppr, border_mode);
84 _copy_array_kernel.configure(&_suppr, update_number, corners, &_num_buffer);
85
86 _suppr.allocator()->allocate();
87 }
88
89 // Allocate intermediate tensors
90 _output.allocator()->allocate();
91}
92
93void CLFastCorners::run()
94{
95 cl::CommandQueue q = CLScheduler::get().queue();
96
97 if(_non_max)
98 {
99 ARM_COMPUTE_ERROR_ON_MSG(_output.cl_buffer().get() == nullptr, "Unconfigured function");
100 const auto out_buffer = static_cast<unsigned char *>(q.enqueueMapBuffer(_output.cl_buffer(), CL_TRUE, CL_MAP_WRITE, 0, _output.info()->total_size()));
101 memset(out_buffer, 0, _output.info()->total_size());
102 q.enqueueUnmapMemObject(_output.cl_buffer(), out_buffer);
103 }
104
105 CLScheduler::get().enqueue(_fast_corners_kernel, false);
106
107 if(_non_max)
108 {
109 _suppr_func.run();
110 }
111
112 CLScheduler::get().enqueue(_copy_array_kernel, false);
113
114 unsigned int get_num_corners = 0;
115 q.enqueueReadBuffer(_num_buffer, CL_TRUE, 0, sizeof(unsigned int), &get_num_corners);
116
117 size_t corner_size = std::min(static_cast<size_t>(get_num_corners), _corners->max_num_values());
118
119 _corners->resize(corner_size);
120
121 if(_num_corners != nullptr)
122 {
123 *_num_corners = get_num_corners;
124 }
125
126 q.flush();
127}