MindSpore Lite is an AI engine that provides AI model inference for different hardware devices. It has been used in a wide range of fields, such as image classification, target recognition, facial recognition, and character recognition.
This document describes the general development process for MindSpore Lite model inference.
## Basic Concepts
Before getting started, you need to understand the following basic concepts:
**Tensor**: a special data structure that is similar to arrays and matrices. It is a basic data structure used in MindSpore Lite network operations.
**Float16 inference**: a mode in which Float16 is used for inference. Float16, also called half-precision, uses 16 bits to represent a number.
## Available APIs
APIs involved in MindSpore Lite model inference are categorized into context APIs, model APIs, and tensor APIs.
### Context APIs
| API | Description |
| ------------------ | ----------------- |
|OH_AI_ContextHandle OH_AI_ContextCreate()|Creates a context object.|
|void OH_AI_ContextSetThreadNum(OH_AI_ContextHandle context, int32_t thread_num)|Sets the number of runtime threads.|
| void OH_AI_ContextSetThreadAffinityMode(OH_AI_ContextHandle context, int mode)|Sets the affinity mode for binding runtime threads to CPU cores, which are categorized into little cores and big cores depending on the CPU frequency.|
|OH_AI_DeviceInfoHandle OH_AI_DeviceInfoCreate(OH_AI_DeviceType device_type)|Creates a runtime device information object.|
|void OH_AI_ContextDestroy(OH_AI_ContextHandle *context)|Destroys a context object.|
|void OH_AI_DeviceInfoSetEnableFP16(OH_AI_DeviceInfoHandle device_info, bool is_fp16)|Sets whether to enable float16 inference. This function is available only for CPU and GPU devices.|
|void OH_AI_ContextAddDeviceInfo(OH_AI_ContextHandle context, OH_AI_DeviceInfoHandle device_info)|Adds a runtime device information object.|
### Model APIs
| API | Description |
| ------------------ | ----------------- |
|OH_AI_ModelHandle OH_AI_ModelCreate()|Creates a model object.|
|OH_AI_Status OH_AI_ModelBuildFromFile(OH_AI_ModelHandle model, const char *model_path,OH_AI_ModelType odel_type, const OH_AI_ContextHandle model_context)|Loads and builds a MindSpore model from a model file.|
|void OH_AI_ModelDestroy(OH_AI_ModelHandle *model)|Destroys a model object.|
### Tensor APIs
| API | Description |
| ------------------ | ----------------- |
|OH_AI_TensorHandleArray OH_AI_ModelGetInputs(const OH_AI_ModelHandle model)|Obtains the input tensor array structure of a model.|
|int64_t OH_AI_TensorGetElementNum(const OH_AI_TensorHandle tensor)|Obtains the number of tensor elements.|
|const char *OH_AI_TensorGetName(const OH_AI_TensorHandle tensor)|Obtains the name of a tensor.|
|OH_AI_DataType OH_AI_TensorGetDataType(const OH_AI_TensorHandle tensor)|Obtains the tensor data type.|
|void *OH_AI_TensorGetMutableData(const OH_AI_TensorHandle tensor)|Obtains the pointer to variable tensor data.|
## How to Develop
The following figure shows the development process for MindSpore Lite model inference.
**Figure 1** Development process for MindSpore Lite model inference
The development process consists of the following main steps:
1. Prepare the required model.
The required model can be downloaded directly or obtained using the model conversion tool.
- If the downloaded model is in the `.ms` format, you can use it directly for inference. The following uses the **mobilenetv2.ms** model as an example.
- If the downloaded model uses a third-party framework, such as TensorFlow, TensorFlow Lite, Caffe, or ONNX, you can use the [model conversion tool](https://www.mindspore.cn/lite/docs/en/r1.5/use/benchmark_tool.html) to convert it to the `.ms` format.
2. Create a context, and set parameters such as the number of runtime threads and device type.
```c
// Create a context, and set the number of runtime threads to 2 and the thread affinity mode to 1 (big cores first).
OH_AI_ContextHandlecontext=OH_AI_ContextCreate();
if(context==NULL){
printf("OH_AI_ContextCreate failed.\n");
returnOH_AI_STATUS_LITE_ERROR;
}
constintthread_num=2;
OH_AI_ContextSetThreadNum(context,thread_num);
OH_AI_ContextSetThreadAffinityMode(context,1);
// Set the device type to CPU, and disable Float16 inference.
If the MindSpore Lite inference framework is no longer needed, you need to destroy the created model.
```c
// Destroy the model.
OH_AI_ModelDestroy(&model);
```
## Verification
1. Compile **CMakeLists.txt**.
```cmake
cmake_minimum_required(VERSION 3.14)
project(Demo)
add_executable(demo main.c)
target_link_libraries(
demo
mindspore-lite.huawei
pthread
dl
)
```
- To use ohos-sdk for cross compilation, you need to set the native toolchain path for the CMake tool as follows: `-DCMAKE_TOOLCHAIN_FILE="/xxx/ohos-sdk/linux/native/build/cmake/ohos.toolchain.cmake"`.
- The toolchain builds a 64-bit application by default. To build a 32-bit application, add the following configuration: `-DOHOS_ARCH="armeabi-v7a"`.
2. Run the CMake tool.
- Use hdc_std to connect to the RK3568 development board and put **demo** and **mobilenetv2.ms** to the same directory on the board.
- Run the hdc_std shell command to access the development board, go to the directory where **demo** is located, and run the following command:
```shell
./demo mobilenetv2.ms
```
The inference is successful if the output is similar to the following:
```shell
# ./QuickStart ./mobilenetv2.ms
Tensor name: Softmax-65, tensor size is 4004 ,elements num: 1001.
Your application can call the **ReminderRequest** class to create scheduled reminders for countdown timers, calendar events, and alarm clocks. When the created reminders are published, the timing and pop-up notification functions of your application will be taken over by the reminder agent in the background, even when your application is frozen or exits.
The **BundleStatusCallback** module provides bundle callback information, which is obtained through [innerBundleManager.on](js-apis-Bundle-InnerBundleManager.md).
> **NOTE**
>
> The initial APIs of this module are supported since API version 8. Newly added APIs will be marked with a superscript to indicate their earliest API version.
## BundleStatusCallback
**System API**: This is a system API and cannot be called by third-party applications.
The **colorSpaceManager** module provides APIs for creating and managing color space objects and obtaining basic color space attributes.
> **NOTE**
>
> The initial APIs of this module are supported since API version 9. Newly added APIs will be marked with a superscript to indicate their earliest API version.
| SRGB | 4 | SRGB.<br>This is the default color space type.|
| CUSTOM | 5 | Custom type.|
## ColorSpacePrimaries
Defines the color space primaries. A color space is defined by chromaticity coordinates of the red, green, and blue additive primaries, the white point, and the gamma.
| colorSpaceName | [ColorSpace](#colorspace)| Yes | Type of the color space.<br>**UNKNOWN** and **CUSTOM** cannot be used when creating standard color space objects. |
**Return value**
| Type | Description |
| ------------------ | ------------------------ |
| [ColorSpaceManager](#colorspacemanager) | Color space object created. |
The Ethernet Connection Management module provides wired network capabilities, which allow users to set the IP address, subnet mask, gateway, and Domain Name System (DNS) server of a wired network.
> **NOTE**
>
> The initial APIs of this module are supported since API version 9. Newly added APIs will be marked with a superscript to indicate their earliest API version.
| iface | string | Yes | Name of the network interface. |
| ic | [InterfaceConfiguration](#interfaceconfiguration) | Yes | Network interface configuration to set. |
| callback | AsyncCallback\<void> | Yes | Callback used to return the result. If the operation is successful, the return result is empty. If the operation fails, an error code is returned.|
| iface | string | No | Name of the network interface. If this parameter is left empty, the API checks for any active network interface. |
| callback | AsyncCallback\<number> | Yes | Callback used to return the result. The value **1** means that the network interface is active, **0** means that the network interface is inactive, and any other value means that an error has occurred.|
| Promise\<number> | Promise used to return the result. The value **1** means that the network interface is active, **0** means that the network interface is inactive, and any other value means that an error has occurred.|
| mode | [IPSetMode](#ipsetmode) | Configuration mode of the Ethernet connection.|
| ipAddr | string | Static IP address of the Ethernet connection. The value must be an IPv4 address, which is a 32-bit number displayed in dotted decimal notation and each 8-bit field ranges from 0 to 255. This parameter does not need to be configured in Dynamic Host Configuration Protocol (DHCP) mode.|
| route | string | Route of the Ethernet connection. The value must be an IPv4 address. This parameter does not need to be configured in DHCP mode.|
| gateway | string | Gateway of the Ethernet connection. The value must be an IPv4 address. This parameter does not need to be configured in DHCP mode.|
| netMask | string | Subnet mask of the Ethernet connection. The value must be an IPv4 address. This parameter does not need to be configured in DHCP mode.|
| dnsServers | string | DNS server addresses of the Ethernet connection. The value must be an IPv4 address. This parameter does not need to be configured in DHCP mode. Multiple addresses are separated by commas (,).|
## IPSetMode
Defines the configuration mode of the Ethernet connection.
The Network Traffic Management module collects statistics on the mobile data traffic and allows you to query the data volume by network interface (cellular or Wi-Fi) or application.
> **NOTE**
>
> The initial APIs of this module are supported since API version 9. Newly added APIs will be marked with a superscript to indicate their earliest API version.
The Transport Layer Security (TLS) protocol is designed to help protect the privacy of information at the transport layer. TLSSocket is an extension to socket communication. It provides higher security than socket communication by adding a security protection layer, which consists of the following submodules: key, certificate, and communication.
> **NOTE**
>
> The initial APIs of this module are supported since API version 9. Newly added APIs will be marked with a superscript to indicate their earliest API version.
Sets up a TLSSocket connection, and creates and initializes a TLS session. During this process, a TLS/SSL handshake is performed between the application and the server to implement data transmission. This API uses an asynchronous callback to return the result.
| options | [TLSConnectOptions](#tlsconnectoptions) | Yes | Parameters required for the connection.|
| callback | AsyncCallback\<void> | Yes | Callback used to return the result. If the operation is successful, the return result is empty. If the operation fails, an error code is returned.|
Sets up a TLSSocket connection, and creates and initializes a TLS session. During this process, a TLS/SSL handshake is performed between the application and the server to implement data transmission. This API uses a promise to return the result.
| Promise\<void> | Promise used to return the result. If the operation is successful, the return result is empty. If the operation fails, an error code is returned.|
Obtains the cipher suites supported by both parties after a TLSSocket connection is established. This API uses an asynchronous callback to return the result.
Obtains the signing algorithms supported by both parties after a TLSSocket connection is established. This API uses an asynchronous callback to return the result.
Your application can call the **ReminderRequest** class to create scheduled reminders for countdown timers, calendar events, and alarm clocks. When the created reminders are published, the timing and pop-up notification functions of your application will be taken over by the reminder agent in the background, even when your application is frozen or exits.
