1# Using MindSpore Lite for Image Classification (C/C++) 2 3## When to Use 4 5You can use [MindSpore](../../reference/apis-mindspore-lite-kit/capi-mindspore.md) to quickly deploy AI algorithms into your application to perform AI model inference for image classification. 6 7Image classification can be used to recognize objects in images and is widely used in medical image analysis, auto driving, e-commerce, and facial recognition. 8 9## Basic Concepts 10 11- N-API: a set of native APIs used to build ArkTS components. N-APIs can be used to encapsulate C/C++ libraries into ArkTS modules. 12 13## Development Process 14 151. Select an image classification model. 162. Use the MindSpore Lite inference model on the device to classify the selected images. 17 18## Environment Setup 19 20Install DevEco Studio 4.1 or later, and update the SDK to API version 11 or later. 21 22## How to Develop 23 24The following uses inference on an image in the album as an example to describe how to use MindSpore Lite to implement image classification. 25 26### Selecting a Model 27 28This sample application uses [mobilenetv2.ms](https://download.mindspore.cn/model_zoo/official/lite/mobilenetv2_openimage_lite/1.5/mobilenetv2.ms) as the image classification model. The model file is available in the **entry/src/main/resources/rawfile** project directory. 29 30If you have other pre-trained models for image classification, convert the original model into the .ms format by referring to [Using MindSpore Lite for Model Conversion](mindspore-lite-converter-guidelines.md). 31 32### Writing Code 33 34#### Image Input and Preprocessing 35 361. Call [@ohos.file.picker](../../reference/apis-core-file-kit/js-apis-file-picker.md) to pick up the desired image in the album. 37 382. Based on the input image size, call [[@ohos.multimedia.image](../../reference/apis-image-kit/arkts-apis-image.md) and [@ohos.file.fs](../../reference/apis-core-file-kit/js-apis-file-fs.md) to perform operations such as cropping the image, obtaining the image buffer, and standardizing the image. 39 40 ```ts 41 // Index.ets 42 import { fileIo } from '@kit.CoreFileKit'; 43 import { photoAccessHelper } from '@kit.MediaLibraryKit'; 44 import { BusinessError } from '@kit.BasicServicesKit'; 45 import { image } from '@kit.ImageKit'; 46 47 @Entry 48 @Component 49 struct Index { 50 @State modelName: string = 'mobilenetv2.ms'; 51 @State modelInputHeight: number = 224; 52 @State modelInputWidth: number = 224; 53 @State uris: Array<string> = []; 54 55 build() { 56 Row() { 57 Column() { 58 Button() { 59 Text('photo') 60 .fontSize(30) 61 .fontWeight(FontWeight.Bold) 62 } 63 .type(ButtonType.Capsule) 64 .margin({ 65 top: 20 66 }) 67 .backgroundColor('#0D9FFB') 68 .width('40%') 69 .height('5%') 70 .onClick(() => { 71 let resMgr = this.getUIContext()?.getHostContext()?.getApplicationContext().resourceManager; 72 73 // Obtain images in an album. 74 // 1. Create an image picker instance. 75 let photoSelectOptions = new photoAccessHelper.PhotoSelectOptions(); 76 77 // 2. Set the media file type to IMAGE and set the maximum number of media files that can be selected. 78 photoSelectOptions.MIMEType = photoAccessHelper.PhotoViewMIMETypes.IMAGE_TYPE; 79 photoSelectOptions.maxSelectNumber = 1; 80 81 // 3. Create an album picker instance and call select() to open the album page for file selection. After file selection is done, the result set is returned through photoSelectResult. 82 let photoPicker = new photoAccessHelper.PhotoViewPicker(); 83 photoPicker.select(photoSelectOptions, 84 async (err: BusinessError, photoSelectResult: photoAccessHelper.PhotoSelectResult) => { 85 if (err) { 86 console.error('MS_LITE_ERR: PhotoViewPicker.select failed with err: ' + JSON.stringify(err)); 87 return; 88 } 89 console.info('MS_LITE_LOG: PhotoViewPicker.select successfully, photoSelectResult uri: ' + 90 JSON.stringify(photoSelectResult)); 91 this.uris = photoSelectResult.photoUris; 92 console.info('MS_LITE_LOG: uri: ' + this.uris); 93 // Preprocess the image data. 94 try { 95 // 1. Based on the specified URI, call fileIo.openSync to open the file to obtain the FD. 96 let file = fileIo.openSync(this.uris[0], fileIo.OpenMode.READ_ONLY); 97 console.info('MS_LITE_LOG: file fd: ' + file.fd); 98 99 // 2. Based on the FD, call fileIo.readSync to read the data in the file. 100 let inputBuffer = new ArrayBuffer(4096000); 101 let readLen = fileIo.readSync(file.fd, inputBuffer); 102 console.info('MS_LITE_LOG: readSync data to file succeed and inputBuffer size is:' + readLen); 103 104 // 3. Perform image preprocessing through PixelMap. 105 let imageSource = image.createImageSource(file.fd); 106 imageSource.createPixelMap().then((pixelMap) => { 107 pixelMap.getImageInfo().then((info) => { 108 console.info('MS_LITE_LOG: info.width = ' + info.size.width); 109 console.info('MS_LITE_LOG: info.height = ' + info.size.height); 110 // 4. Crop the image based on the input image size and obtain the image buffer readBuffer. 111 pixelMap.scale(256.0 / info.size.width, 256.0 / info.size.height).then(() => { 112 pixelMap.crop({ 113 x: 16, 114 y: 16, 115 size: { height: this.modelInputHeight, width: this.modelInputWidth } 116 }) 117 .then(async () => { 118 let info = await pixelMap.getImageInfo(); 119 console.info('MS_LITE_LOG: crop info.width = ' + info.size.width); 120 console.info('MS_LITE_LOG: crop info.height = ' + info.size.height); 121 // Set the size of readBuffer. 122 let readBuffer = new ArrayBuffer(this.modelInputHeight * this.modelInputWidth * 4); 123 await pixelMap.readPixelsToBuffer(readBuffer); 124 console.info('MS_LITE_LOG: Succeeded in reading image pixel data, buffer: ' + 125 readBuffer.byteLength); 126 // Convert readBuffer to the float32 format, and standardize the image. 127 const imageArr = 128 new Uint8Array(readBuffer.slice(0, this.modelInputHeight * this.modelInputWidth * 4)); 129 console.info('MS_LITE_LOG: imageArr length: ' + imageArr.length); 130 let means = [0.485, 0.456, 0.406]; 131 let stds = [0.229, 0.224, 0.225]; 132 let float32View = new Float32Array(this.modelInputHeight * this.modelInputWidth * 3); 133 let index = 0; 134 for (let i = 0; i < imageArr.length; i++) { 135 if ((i + 1) % 4 == 0) { 136 float32View[index] = (imageArr[i - 3] / 255.0 - means[0]) / stds[0]; // B 137 float32View[index+1] = (imageArr[i - 2] / 255.0 - means[1]) / stds[1]; // G 138 float32View[index+2] = (imageArr[i - 1] / 255.0 - means[2]) / stds[2]; // R 139 index += 3; 140 } 141 } 142 console.info('MS_LITE_LOG: float32View length: ' + float32View.length); 143 let printStr = 'float32View data:'; 144 for (let i = 0; i < 20; i++) { 145 printStr += ' ' + float32View[i]; 146 } 147 console.info('MS_LITE_LOG: float32View data: ' + printStr); 148 }) 149 }) 150 }) 151 }) 152 } catch (err) { 153 console.error('MS_LITE_LOG: uri: open file fd failed.' + err); 154 } 155 }) 156 }) 157 }.width('100%') 158 } 159 .height('100%') 160 } 161 } 162 ``` 163 164#### Writing Inference Code 165 166Call [MindSpore](../../reference/apis-mindspore-lite-kit/capi-mindspore.md) to implement inference on the device. The operation process is as follows: 167 1681. Include the corresponding header file. 169 170 ```c++ 171 #include <iostream> 172 #include <sstream> 173 #include <stdlib.h> 174 #include <hilog/log.h> 175 #include <rawfile/raw_file_manager.h> 176 #include <mindspore/types.h> 177 #include <mindspore/model.h> 178 #include <mindspore/context.h> 179 #include <mindspore/status.h> 180 #include <mindspore/tensor.h> 181 #include "napi/native_api.h" 182 ``` 183 1842. Read the model file. 185 186 ```c++ 187 #define LOGI(...) ((void)OH_LOG_Print(LOG_APP, LOG_INFO, LOG_DOMAIN, "[MSLiteNapi]", __VA_ARGS__)) 188 #define LOGD(...) ((void)OH_LOG_Print(LOG_APP, LOG_DEBUG, LOG_DOMAIN, "[MSLiteNapi]", __VA_ARGS__)) 189 #define LOGW(...) ((void)OH_LOG_Print(LOG_APP, LOG_WARN, LOG_DOMAIN, "[MSLiteNapi]", __VA_ARGS__)) 190 #define LOGE(...) ((void)OH_LOG_Print(LOG_APP, LOG_ERROR, LOG_DOMAIN, "[MSLiteNapi]", __VA_ARGS__)) 191 192 void *ReadModelFile(NativeResourceManager *nativeResourceManager, const std::string &modelName, size_t *modelSize) { 193 auto rawFile = OH_ResourceManager_OpenRawFile(nativeResourceManager, modelName.c_str()); 194 if (rawFile == nullptr) { 195 LOGE("MS_LITE_ERR: Open model file failed"); 196 return nullptr; 197 } 198 long fileSize = OH_ResourceManager_GetRawFileSize(rawFile); 199 void *modelBuffer = malloc(fileSize); 200 if (modelBuffer == nullptr) { 201 LOGE("MS_LITE_ERR: OH_ResourceManager_ReadRawFile failed"); 202 } 203 int ret = OH_ResourceManager_ReadRawFile(rawFile, modelBuffer, fileSize); 204 if (ret == 0) { 205 LOGI("MS_LITE_LOG: OH_ResourceManager_ReadRawFile failed"); 206 OH_ResourceManager_CloseRawFile(rawFile); 207 return nullptr; 208 } 209 OH_ResourceManager_CloseRawFile(rawFile); 210 *modelSize = fileSize; 211 return modelBuffer; 212 } 213 ``` 214 2153. Create a context, set parameters such as the number of threads and device type, and load the model. The sample model does not support NNRt inference. 216 217 ```c++ 218 void DestroyModelBuffer(void **buffer) { 219 if (buffer == nullptr) { 220 return; 221 } 222 free(*buffer); 223 *buffer = nullptr; 224 } 225 226 OH_AI_ContextHandle CreateMSLiteContext(void *modelBuffer) { 227 // Set executing context for model. 228 auto context = OH_AI_ContextCreate(); 229 if (context == nullptr) { 230 DestroyModelBuffer(&modelBuffer); 231 LOGE("MS_LITE_ERR: Create MSLite context failed.\n"); 232 return nullptr; 233 } 234 // The OH_AI_DeviceInfoCreate(OH_AI_DEVICETYPE_NNRT) option is not supported. 235 auto cpu_device_info = OH_AI_DeviceInfoCreate(OH_AI_DEVICETYPE_CPU); 236 237 OH_AI_DeviceInfoSetEnableFP16(cpu_device_info, true); 238 OH_AI_ContextAddDeviceInfo(context, cpu_device_info); 239 240 LOGI("MS_LITE_LOG: Build MSLite context success.\n"); 241 return context; 242 } 243 244 OH_AI_ModelHandle CreateMSLiteModel(void *modelBuffer, size_t modelSize, OH_AI_ContextHandle context) { 245 // Create model 246 auto model = OH_AI_ModelCreate(); 247 if (model == nullptr) { 248 DestroyModelBuffer(&modelBuffer); 249 LOGE("MS_LITE_ERR: Allocate MSLite Model failed.\n"); 250 return nullptr; 251 } 252 253 // Build model object 254 auto build_ret = OH_AI_ModelBuild(model, modelBuffer, modelSize, OH_AI_MODELTYPE_MINDIR, context); 255 DestroyModelBuffer(&modelBuffer); 256 if (build_ret != OH_AI_STATUS_SUCCESS) { 257 OH_AI_ModelDestroy(&model); 258 LOGE("MS_LITE_ERR: Build MSLite model failed.\n"); 259 return nullptr; 260 } 261 LOGI("MS_LITE_LOG: Build MSLite model success.\n"); 262 return model; 263 } 264 ``` 265 2664. Set the model input data and perform model inference. 267 268 ```c++ 269 constexpr int K_NUM_PRINT_OF_OUT_DATA = 20; 270 271 // Set the model input data. 272 int FillInputTensor(OH_AI_TensorHandle input, std::vector<float> input_data) { 273 if (OH_AI_TensorGetDataType(input) == OH_AI_DATATYPE_NUMBERTYPE_FLOAT32) { 274 float *data = (float *)OH_AI_TensorGetMutableData(input); 275 for (size_t i = 0; i < OH_AI_TensorGetElementNum(input); i++) { 276 data[i] = input_data[i]; 277 } 278 return OH_AI_STATUS_SUCCESS; 279 } else { 280 return OH_AI_STATUS_LITE_ERROR; 281 } 282 } 283 284 // Execute model inference. 285 int RunMSLiteModel(OH_AI_ModelHandle model, std::vector<float> input_data) { 286 // Set input data for model. 287 auto inputs = OH_AI_ModelGetInputs(model); 288 289 auto ret = FillInputTensor(inputs.handle_list[0], input_data); 290 if (ret != OH_AI_STATUS_SUCCESS) { 291 LOGE("MS_LITE_ERR: RunMSLiteModel set input error.\n"); 292 return OH_AI_STATUS_LITE_ERROR; 293 } 294 // Get model output. 295 auto outputs = OH_AI_ModelGetOutputs(model); 296 // Predict model. 297 auto predict_ret = OH_AI_ModelPredict(model, inputs, &outputs, nullptr, nullptr); 298 if (predict_ret != OH_AI_STATUS_SUCCESS) { 299 LOGE("MS_LITE_ERR: MSLite Predict error.\n"); 300 return OH_AI_STATUS_LITE_ERROR; 301 } 302 LOGI("MS_LITE_LOG: Run MSLite model Predict success.\n"); 303 // Print output tensor data. 304 LOGI("MS_LITE_LOG: Get model outputs:\n"); 305 for (size_t i = 0; i < outputs.handle_num; i++) { 306 auto tensor = outputs.handle_list[i]; 307 LOGI("MS_LITE_LOG: - Tensor %{public}d name is: %{public}s.\n", static_cast<int>(i), 308 OH_AI_TensorGetName(tensor)); 309 LOGI("MS_LITE_LOG: - Tensor %{public}d size is: %{public}d.\n", static_cast<int>(i), 310 (int)OH_AI_TensorGetDataSize(tensor)); 311 LOGI("MS_LITE_LOG: - Tensor data is:\n"); 312 auto out_data = reinterpret_cast<const float *>(OH_AI_TensorGetData(tensor)); 313 std::stringstream outStr; 314 for (int i = 0; (i < OH_AI_TensorGetElementNum(tensor)) && (i <= K_NUM_PRINT_OF_OUT_DATA); i++) { 315 outStr << out_data[i] << " "; 316 } 317 LOGI("MS_LITE_LOG: %{public}s", outStr.str().c_str()); 318 } 319 return OH_AI_STATUS_SUCCESS; 320 } 321 ``` 322 3235. Implement a complete model inference process. 324 325 ```c++ 326 static napi_value RunDemo(napi_env env, napi_callback_info info) { 327 LOGI("MS_LITE_LOG: Enter runDemo()"); 328 napi_value error_ret; 329 napi_create_int32(env, -1, &error_ret); 330 // Process the input data. 331 size_t argc = 2; 332 napi_value argv[2] = {nullptr}; 333 napi_get_cb_info(env, info, &argc, argv, nullptr, nullptr); 334 bool isArray = false; 335 napi_is_array(env, argv[0], &isArray); 336 uint32_t length = 0; 337 // Obtain the length of the array. 338 napi_get_array_length(env, argv[0], &length); 339 LOGI("MS_LITE_LOG: argv array length = %{public}d", length); 340 std::vector<float> input_data; 341 double param = 0; 342 for (int i = 0; i < length; i++) { 343 napi_value value; 344 napi_get_element(env, argv[0], i, &value); 345 napi_get_value_double(env, value, ¶m); 346 input_data.push_back(static_cast<float>(param)); 347 } 348 std::stringstream outstr; 349 for (int i = 0; i < K_NUM_PRINT_OF_OUT_DATA; i++) { 350 outstr << input_data[i] << " "; 351 } 352 LOGI("MS_LITE_LOG: input_data = %{public}s", outstr.str().c_str()); 353 // Read model file 354 const std::string modelName = "mobilenetv2.ms"; 355 LOGI("MS_LITE_LOG: Run model: %{public}s", modelName.c_str()); 356 size_t modelSize; 357 auto resourcesManager = OH_ResourceManager_InitNativeResourceManager(env, argv[1]); 358 auto modelBuffer = ReadModelFile(resourcesManager, modelName, &modelSize); 359 if (modelBuffer == nullptr) { 360 LOGE("MS_LITE_ERR: Read model failed"); 361 return error_ret; 362 } 363 LOGI("MS_LITE_LOG: Read model file success"); 364 365 auto context = CreateMSLiteContext(modelBuffer); 366 if (context == nullptr) { 367 LOGE("MS_LITE_ERR: MSLiteFwk Build context failed.\n"); 368 return error_ret; 369 } 370 auto model = CreateMSLiteModel(modelBuffer, modelSize, context); 371 if (model == nullptr) { 372 OH_AI_ContextDestroy(&context); 373 LOGE("MS_LITE_ERR: MSLiteFwk Build model failed.\n"); 374 return error_ret; 375 } 376 int ret = RunMSLiteModel(model, input_data); 377 if (ret != OH_AI_STATUS_SUCCESS) { 378 OH_AI_ModelDestroy(&model); 379 OH_AI_ContextDestroy(&context); 380 LOGE("MS_LITE_ERR: RunMSLiteModel failed.\n"); 381 return error_ret; 382 } 383 napi_value out_data; 384 napi_create_array(env, &out_data); 385 auto outputs = OH_AI_ModelGetOutputs(model); 386 OH_AI_TensorHandle output_0 = outputs.handle_list[0]; 387 float *output0Data = reinterpret_cast<float *>(OH_AI_TensorGetMutableData(output_0)); 388 for (size_t i = 0; i < OH_AI_TensorGetElementNum(output_0); i++) { 389 napi_value element; 390 napi_create_double(env, static_cast<double>(output0Data[i]), &element); 391 napi_set_element(env, out_data, i, element); 392 } 393 OH_AI_ModelDestroy(&model); 394 OH_AI_ContextDestroy(&context); 395 LOGI("MS_LITE_LOG: Exit runDemo()"); 396 return out_data; 397 } 398 ``` 399 4006. Write the **CMake** script to link the MindSpore Lite dynamic library. 401 402 ```c++ 403 # the minimum version of CMake. 404 cmake_minimum_required(VERSION 3.4.1) 405 project(MindSporeLiteCDemo) 406 407 set(NATIVERENDER_ROOT_PATH ${CMAKE_CURRENT_SOURCE_DIR}) 408 409 if(DEFINED PACKAGE_FIND_FILE) 410 include(${PACKAGE_FIND_FILE}) 411 endif() 412 413 include_directories(${NATIVERENDER_ROOT_PATH} 414 ${NATIVERENDER_ROOT_PATH}/include) 415 416 add_library(entry SHARED mslite_napi.cpp) 417 target_link_libraries(entry PUBLIC mindspore_lite_ndk) 418 target_link_libraries(entry PUBLIC hilog_ndk.z) 419 target_link_libraries(entry PUBLIC rawfile.z) 420 target_link_libraries(entry PUBLIC ace_napi.z) 421 ``` 422 423#### Use N-APIs to encapsulate the C++ dynamic library into an ArkTS module. 424 4251. In **entry/src/main/cpp/types/libentry/Index.d.ts**, define the ArkTS API **runDemo ()**. The content is as follows: 426 427 ```ts 428 export const runDemo: (a: number[], b:Object) => Array<number>; 429 ``` 430 4312. In the **oh-package.json5** file, associate the API with the .so file to form a complete ArkTS module. 432 433 ```json 434 { 435 "name": "libentry.so", 436 "types": "./Index.d.ts", 437 "version": "1.0.0", 438 "description": "MindSpore Lite inference module" 439 } 440 ``` 441 442#### Invoke the encapsulated ArkTS module for inference and output the result. 443 444In **entry/src/main/ets/pages/Index.ets**, call the encapsulated ArkTS module to process the inference result. 445 446```ts 447// Index.ets 448import msliteNapi from 'libentry.so' 449 450@Entry 451@Component 452struct Index { 453 @State modelInputHeight: number = 224; 454 @State modelInputWidth: number = 224; 455 @State max: number = 0; 456 @State maxIndex: number = 0; 457 @State maxArray: Array<number> = []; 458 @State maxIndexArray: Array<number> = []; 459 460 build() { 461 Row() { 462 Column() { 463 Button() { 464 Text('photo') 465 .fontSize(30) 466 .fontWeight(FontWeight.Bold) 467 } 468 .type(ButtonType.Capsule) 469 .margin({ 470 top: 20 471 }) 472 .backgroundColor('#0D9FFB') 473 .width('40%') 474 .height('5%') 475 .onClick(() => { 476 let resMgr = this.getUIContext()?.getHostContext()?.getApplicationContext().resourceManager; 477 let float32View = new Float32Array(this.modelInputHeight * this.modelInputWidth * 3); 478 // Image input and preprocessing 479 // Call the C++ runDemo function. The buffer data of the input image is stored in float32View after preprocessing. For details, see Image Input and Preprocessing. 480 console.info('MS_LITE_LOG: *** Start MSLite Demo ***'); 481 let output: Array<number> = msliteNapi.runDemo(Array.from(float32View), resMgr); 482 483 // Obtain the maximum number of categories. 484 this.max = 0; 485 this.maxIndex = 0; 486 this.maxArray = []; 487 this.maxIndexArray = []; 488 let newArray = output.filter(value => value !== this.max); 489 for (let n = 0; n < 5; n++) { 490 this.max = output[0]; 491 this.maxIndex = 0; 492 for (let m = 0; m < newArray.length; m++) { 493 if (newArray[m] > this.max) { 494 this.max = newArray[m]; 495 this.maxIndex = m; 496 } 497 } 498 this.maxArray.push(Math.round(this.max * 10000)); 499 this.maxIndexArray.push(this.maxIndex); 500 // Call the array filter function. 501 newArray = newArray.filter(value => value !== this.max); 502 } 503 console.info('MS_LITE_LOG: max:' + this.maxArray); 504 console.info('MS_LITE_LOG: maxIndex:' + this.maxIndexArray); 505 console.info('MS_LITE_LOG: *** Finished MSLite Demo ***'); 506 }) 507 }.width('100%') 508 } 509 .height('100%') 510 } 511} 512``` 513 514### Debugging and Verification 515 5161. On DevEco Studio, connect to the device, click **Run entry**, and build your own HAP. 517 518 ```shell 519 Launching com.samples.mindsporelitecdemo 520 $ hdc shell aa force-stop com.samples.mindsporelitecdemo 521 $ hdc shell mkdir data/local/tmp/xxx 522 $ hdc file send C:\Users\xxx\MindSporeLiteCDemo\entry\build\default\outputs\default\entry-default-signed.hap "data/local/tmp/xxx" 523 $ hdc shell bm install -p data/local/tmp/xxx 524 $ hdc shell rm -rf data/local/tmp/xxx 525 $ hdc shell aa start -a EntryAbility -b com.samples.mindsporelitecdemo 526 ``` 527 5282. Touch the **photo** button on the device screen, select an image, and touch **OK**. The classification result of the selected image is displayed on the device screen. In the log printing result, filter images by the keyword **MS_LITE**. The following information is displayed: 529 530 ```verilog 531 08-05 17:15:52.001 4684-4684 A03d00/JSAPP pid-4684 I MS_LITE_LOG: PhotoViewPicker.select successfully, photoSelectResult uri: {"photoUris":["file://media/Photo/13/IMG_1501955351_012/plant.jpg"]} 532 ... 533 08-05 17:15:52.627 4684-4684 A03d00/JSAPP pid-4684 I MS_LITE_LOG: crop info.width = 224 534 08-05 17:15:52.627 4684-4684 A03d00/JSAPP pid-4684 I MS_LITE_LOG: crop info.height = 224 535 08-05 17:15:52.628 4684-4684 A03d00/JSAPP pid-4684 I MS_LITE_LOG: Succeeded in reading image pixel data, buffer: 200704 536 08-05 17:15:52.971 4684-4684 A03d00/JSAPP pid-4684 I MS_LITE_LOG: float32View data: float32View data: 1.2385478019714355 1.308123230934143 1.4722440242767334 1.2385478019714355 1.308123230934143 1.4722440242767334 1.2385478019714355 1.308123230934143 1.4722440242767334 1.2385478019714355 1.308123230934143 1.4722440242767334 1.2385478019714355 1.308123230934143 1.4722440242767334 1.2385478019714355 1.308123230934143 1.4722440242767334 1.2385478019714355 1.308123230934143 537 08-05 17:15:52.971 4684-4684 A03d00/JSAPP pid-4684 I MS_LITE_LOG: *** Start MSLite Demo *** 538 08-05 17:15:53.454 4684-4684 A00000/[MSLiteNapi] pid-4684 I MS_LITE_LOG: Build MSLite model success. 539 08-05 17:15:53.753 4684-4684 A00000/[MSLiteNapi] pid-4684 I MS_LITE_LOG: Run MSLite model Predict success. 540 08-05 17:15:53.753 4684-4684 A00000/[MSLiteNapi] pid-4684 I MS_LITE_LOG: Get model outputs: 541 08-05 17:15:53.753 4684-4684 A00000/[MSLiteNapi] pid-4684 I MS_LITE_LOG: - Tensor 0 name is: Default/head-MobileNetV2Head/Sigmoid-op466. 542 08-05 17:15:53.753 4684-4684 A00000/[MSLiteNapi] pid-4684 I MS_LITE_LOG: - Tensor data is: 543 08-05 17:15:53.753 4684-4684 A00000/[MSLiteNapi] pid-4684 I MS_LITE_LOG: 3.43385e-06 1.40285e-05 9.11969e-07 4.91007e-05 9.50266e-07 3.94537e-07 0.0434676 3.97196e-05 0.00054832 0.000246202 1.576e-05 3.6494e-06 1.23553e-05 0.196977 5.3028e-05 3.29346e-05 4.90475e-07 1.66109e-06 7.03273e-06 8.83677e-07 3.1365e-06 544 08-05 17:15:53.781 4684-4684 A03d00/JSAPP pid-4684 W MS_LITE_WARN: output length = 500 ;value = 0.0000034338463592575863,0.000014028532859811094,9.119685273617506e-7,0.000049100715841632336,9.502661555416125e-7,3.945370394831116e-7,0.04346757382154465,0.00003971960904891603,0.0005483203567564487,0.00024620210751891136,0.000015759984307806008,0.0000036493988773145247,0.00001235533181898063,0.1969769448041916,0.000053027983085485175,0.000032934600312728435,4.904751449430478e-7,0.0000016610861166554969,0.000007032729172351537,8.836767619868624e-7 545 08-05 17:15:53.831 4684-4684 A03d00/JSAPP pid-4684 I MS_LITE_LOG: max:9497,7756,1970,435,46 546 08-05 17:15:53.831 4684-4684 A03d00/JSAPP pid-4684 I MS_LITE_LOG: maxIndex:323,46,13,6,349 547 08-05 17:15:53.831 4684-4684 A03d00/JSAPP pid-4684 I MS_LITE_LOG: *** Finished MSLite Demo *** 548 ``` 549 550 551### Effects 552 553Touch the **photo** button on the device screen, select an image, and touch **OK**. The top 4 categories of the image are displayed below the image. 554 555  556 557  558 559 560
