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
|
/*
* Copyright (c) 2021-2022 Arm Limited. All rights reserved.
* SPDX-License-Identifier: Apache-2.0
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#include "UseCaseHandler.hpp"
#include "VisualWakeWordModel.hpp"
#include "Classifier.hpp"
#include "InputFiles.hpp"
#include "ImageUtils.hpp"
#include "UseCaseCommonUtils.hpp"
#include "hal.h"
#include "log_macros.h"
#include "VisualWakeWordProcessing.hpp"
namespace arm {
namespace app {
/* Visual Wake Word inference handler. */
bool ClassifyImageHandler(ApplicationContext &ctx, uint32_t imgIndex, bool runAll)
{
auto& profiler = ctx.Get<Profiler&>("profiler");
auto& model = ctx.Get<Model&>("model");
/* If the request has a valid size, set the image index. */
if (imgIndex < NUMBER_OF_FILES) {
if (!SetAppCtxIfmIdx(ctx, imgIndex,"imgIndex")) {
return false;
}
}
auto initialImgIdx = ctx.Get<uint32_t>("imgIndex");
constexpr uint32_t dataPsnImgDownscaleFactor = 1;
constexpr uint32_t dataPsnImgStartX = 10;
constexpr uint32_t dataPsnImgStartY = 35;
constexpr uint32_t dataPsnTxtInfStartX = 150;
constexpr uint32_t dataPsnTxtInfStartY = 70;
if (!model.IsInited()) {
printf_err("Model is not initialised! Terminating processing.\n");
return false;
}
TfLiteTensor* inputTensor = model.GetInputTensor(0);
TfLiteTensor* outputTensor = model.GetOutputTensor(0);
if (!inputTensor->dims) {
printf_err("Invalid input tensor dims\n");
return false;
} else if (inputTensor->dims->size < 4) {
printf_err("Input tensor dimension should be = 4\n");
return false;
}
/* Get input shape for displaying the image. */
TfLiteIntArray* inputShape = model.GetInputShape(0);
const uint32_t nCols = inputShape->data[arm::app::VisualWakeWordModel::ms_inputColsIdx];
const uint32_t nRows = inputShape->data[arm::app::VisualWakeWordModel::ms_inputRowsIdx];
if (arm::app::VisualWakeWordModel::ms_inputChannelsIdx >= static_cast<uint32_t>(inputShape->size)) {
printf_err("Invalid channel index.\n");
return false;
}
/* We expect RGB images to be provided. */
const uint32_t displayChannels = 3;
/* Set up pre and post-processing. */
VisualWakeWordPreProcess preProcess = VisualWakeWordPreProcess(inputTensor);
std::vector<ClassificationResult> results;
VisualWakeWordPostProcess postProcess = VisualWakeWordPostProcess(outputTensor,
ctx.Get<Classifier&>("classifier"),
ctx.Get<std::vector<std::string>&>("labels"), results);
do {
hal_lcd_clear(COLOR_BLACK);
/* Strings for presentation/logging. */
std::string str_inf{"Running inference... "};
const uint8_t* imgSrc = get_img_array(ctx.Get<uint32_t>("imgIndex"));
if (nullptr == imgSrc) {
printf_err("Failed to get image index %" PRIu32 " (max: %u)\n", ctx.Get<uint32_t>("imgIndex"),
NUMBER_OF_FILES - 1);
return false;
}
/* Display this image on the LCD. */
hal_lcd_display_image(
imgSrc,
nCols, nRows, displayChannels,
dataPsnImgStartX, dataPsnImgStartY, dataPsnImgDownscaleFactor);
/* Display message on the LCD - inference running. */
hal_lcd_display_text(str_inf.c_str(), str_inf.size(),
dataPsnTxtInfStartX, dataPsnTxtInfStartY, 0);
/* Run inference over this image. */
info("Running inference on image %" PRIu32 " => %s\n", ctx.Get<uint32_t>("imgIndex"),
get_filename(ctx.Get<uint32_t>("imgIndex")));
const size_t imgSz = inputTensor->bytes < IMAGE_DATA_SIZE ?
inputTensor->bytes : IMAGE_DATA_SIZE;
/* Run the pre-processing, inference and post-processing. */
if (!preProcess.DoPreProcess(imgSrc, imgSz)) {
printf_err("Pre-processing failed.");
return false;
}
if (!RunInference(model, profiler)) {
printf_err("Inference failed.");
return false;
}
if (!postProcess.DoPostProcess()) {
printf_err("Post-processing failed.");
return false;
}
/* Erase. */
str_inf = std::string(str_inf.size(), ' ');
hal_lcd_display_text(str_inf.c_str(), str_inf.size(),
dataPsnTxtInfStartX, dataPsnTxtInfStartY, 0);
/* Add results to context for access outside handler. */
ctx.Set<std::vector<ClassificationResult>>("results", results);
#if VERIFY_TEST_OUTPUT
arm::app::DumpTensor(outputTensor);
#endif /* VERIFY_TEST_OUTPUT */
if (!PresentInferenceResult(results)) {
return false;
}
profiler.PrintProfilingResult();
IncrementAppCtxIfmIdx(ctx,"imgIndex");
} while (runAll && ctx.Get<uint32_t>("imgIndex") != initialImgIdx);
return true;
}
} /* namespace app */
} /* namespace arm */
|
