Richard Burton | ec5e99b | 2022-10-05 11:00:37 +0100 | [diff] [blame] | 1 | /* |
| 2 | * Copyright (c) 2022 Arm Limited. All rights reserved. |
| 3 | * SPDX-License-Identifier: Apache-2.0 |
| 4 | * |
| 5 | * Licensed under the Apache License, Version 2.0 (the "License"); |
| 6 | * you may not use this file except in compliance with the License. |
| 7 | * You may obtain a copy of the License at |
| 8 | * |
| 9 | * http://www.apache.org/licenses/LICENSE-2.0 |
| 10 | * |
| 11 | * Unless required by applicable law or agreed to in writing, software |
| 12 | * distributed under the License is distributed on an "AS IS" BASIS, |
| 13 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 14 | * See the License for the specific language governing permissions and |
| 15 | * limitations under the License. |
| 16 | */ |
| 17 | #include "KwsClassifier.hpp" |
| 18 | |
| 19 | #include <catch.hpp> |
| 20 | |
| 21 | TEST_CASE("Test invalid classifier") |
| 22 | { |
| 23 | TfLiteTensor* outputTens = nullptr; |
| 24 | std::vector<arm::app::ClassificationResult> resultVec; |
| 25 | arm::app::KwsClassifier classifier; |
| 26 | std::vector<std::vector<float>> resultHistory; |
| 27 | REQUIRE(!classifier.GetClassificationResults(outputTens, resultVec, {}, 5, true, resultHistory)); |
| 28 | } |
| 29 | |
| 30 | TEST_CASE("Test valid classifier, average=0 should be same as 1)") |
| 31 | { |
| 32 | int dimArray[] = {1, 5}; |
| 33 | std::vector<std::string> labels(5); |
| 34 | std::vector<uint8_t> outputVec = {0, 1, 2, 3, 4}; |
| 35 | std::vector<std::vector<float>> resultHistory = {}; |
| 36 | TfLiteIntArray* dims= tflite::testing::IntArrayFromInts(dimArray); |
| 37 | TfLiteTensor tfTensor = tflite::testing::CreateQuantizedTensor( |
| 38 | outputVec.data(), dims, 1, 0); |
| 39 | TfLiteTensor* outputTensor = &tfTensor; |
| 40 | std::vector<arm::app::ClassificationResult> resultVec; |
| 41 | arm::app::KwsClassifier classifier; |
| 42 | |
| 43 | REQUIRE(classifier.GetClassificationResults(outputTensor, resultVec, labels, 1, false, resultHistory)); |
| 44 | REQUIRE(resultVec[0].m_labelIdx == 4); |
| 45 | REQUIRE(resultVec[0].m_normalisedVal == 4); |
| 46 | |
| 47 | REQUIRE(classifier.GetClassificationResults(outputTensor, resultVec, labels, 1, false, resultHistory)); |
| 48 | REQUIRE(resultVec[0].m_labelIdx == 4); |
| 49 | REQUIRE(resultVec[0].m_normalisedVal == 4); |
| 50 | |
| 51 | std::vector<std::vector<float>> expectedHistory = {}; |
| 52 | REQUIRE(resultHistory == expectedHistory); |
| 53 | } |
| 54 | |
| 55 | TEST_CASE("Test valid classifier UINT8, average=1, softmax=false") |
| 56 | { |
| 57 | int dimArray[] = {1, 5}; |
| 58 | std::vector<std::string> labels(5); |
| 59 | std::vector<uint8_t> outputVec = {0, 1, 2, 3, 4}; |
| 60 | std::vector<std::vector<float>> resultHistory = {{0, 0, 0, 0, 0}}; |
| 61 | TfLiteIntArray* dims= tflite::testing::IntArrayFromInts(dimArray); |
| 62 | TfLiteTensor tfTensor = tflite::testing::CreateQuantizedTensor( |
| 63 | outputVec.data(), dims, 1, 0); |
| 64 | TfLiteTensor* outputTensor = &tfTensor; |
| 65 | std::vector<arm::app::ClassificationResult> resultVec; |
| 66 | arm::app::KwsClassifier classifier; |
| 67 | |
| 68 | REQUIRE(classifier.GetClassificationResults(outputTensor, resultVec, labels, 1, false, resultHistory)); |
| 69 | REQUIRE(resultVec[0].m_labelIdx == 4); |
| 70 | REQUIRE(resultVec[0].m_normalisedVal == 4); |
| 71 | |
| 72 | /* We do not update history if not >1 in size. */ |
| 73 | std::vector<std::vector<float>> expectedHistory = {{0, 0, 0, 0, 0}}; |
| 74 | REQUIRE(resultHistory == expectedHistory); |
| 75 | } |
| 76 | |
| 77 | TEST_CASE("Test valid classifier UINT8, average=2") |
| 78 | { |
| 79 | int dimArray[] = {1, 5}; |
| 80 | std::vector<std::string> labels(5); |
| 81 | std::vector<uint8_t> outputVec = {0, 1, 2, 3, 4}; |
| 82 | std::vector<std::vector<float>> resultHistory = {{0, 0, 0, 0, 0}, {0, 0, 0, 0, 0}}; |
| 83 | TfLiteIntArray* dims= tflite::testing::IntArrayFromInts(dimArray); |
| 84 | TfLiteTensor tfTensor = tflite::testing::CreateQuantizedTensor( |
| 85 | outputVec.data(), dims, 1, 0); |
| 86 | TfLiteTensor* outputTensor = &tfTensor; |
| 87 | std::vector<arm::app::ClassificationResult> resultVec; |
| 88 | arm::app::KwsClassifier classifier; |
| 89 | |
| 90 | REQUIRE(classifier.GetClassificationResults(outputTensor, resultVec, labels, 1, false, resultHistory)); |
| 91 | REQUIRE(resultVec[0].m_labelIdx == 4); |
| 92 | REQUIRE(resultVec[0].m_normalisedVal == 2); |
| 93 | |
| 94 | REQUIRE(classifier.GetClassificationResults(outputTensor, resultVec, labels, 1, false, resultHistory)); |
| 95 | REQUIRE(resultVec[0].m_labelIdx == 4); |
| 96 | REQUIRE(resultVec[0].m_normalisedVal == 4); |
| 97 | |
| 98 | std::vector<std::vector<float>> expectedHistory = {{0, 1, 2, 3, 4}, {0, 1, 2, 3, 4}}; |
| 99 | REQUIRE(resultHistory == expectedHistory); |
| 100 | } |
| 101 | |
| 102 | TEST_CASE("Test valid classifier int8, average=0") |
| 103 | { |
| 104 | int dimArray[] = {1, 5}; |
| 105 | std::vector<std::string> labels(5); |
| 106 | std::vector<int8_t> outputVec = {-2, -1, 0, 2, 1}; |
| 107 | std::vector<std::vector<float>> resultHistory = {}; |
| 108 | TfLiteIntArray* dims= tflite::testing::IntArrayFromInts(dimArray); |
| 109 | TfLiteTensor tfTensor = tflite::testing::CreateQuantizedTensor( |
| 110 | outputVec.data(), dims, 1, 0); |
| 111 | TfLiteTensor* outputTensor = &tfTensor; |
| 112 | std::vector<arm::app::ClassificationResult> resultVec; |
| 113 | arm::app::KwsClassifier classifier; |
| 114 | |
| 115 | REQUIRE(classifier.GetClassificationResults(outputTensor, resultVec, labels, 1, false, resultHistory)); |
| 116 | REQUIRE(resultVec[0].m_labelIdx == 3); |
| 117 | REQUIRE(resultVec[0].m_normalisedVal == 2); |
| 118 | |
| 119 | REQUIRE(classifier.GetClassificationResults(outputTensor, resultVec, labels, 1, false, resultHistory)); |
| 120 | REQUIRE(resultVec[0].m_labelIdx == 3); |
| 121 | REQUIRE(resultVec[0].m_normalisedVal == 2); |
| 122 | |
| 123 | std::vector<std::vector<float>> expectedHistory = {}; |
| 124 | REQUIRE(resultHistory == expectedHistory); |
| 125 | } |