如何在STM32中利用CubeAI部署自己的神经网络模型（以手写数字识别为例）

def create_mnist_model():    print("TensorFlow版本:", tf.__version__)
    # 加载MNIST数据集    (train_images, train_labels), (test_images, test_labels) = tf.keras.datasets.mnist.load_data()
    # 数据预处理：将图像调整为40x40    def resize_images(images):        resized = []        for img in images:            img = tf.image.resize(img[..., tf.newaxis], [40, 40]).numpy()            resized.append(img)        return np.array(resized)
    train_images = resize_images(train_images)    test_images = resize_images(test_images)
    train_images = train_images.astype('float32') / 255    test_images = test_images.astype('float32') / 255
    # 创建模型    model = models.Sequential([        layers.Conv2D(32, (3, 3), activation='relu', input_shape=(40, 40, 1)),        layers.MaxPooling2D((2, 2)),        layers.Conv2D(64, (3, 3), activation='relu'),        layers.MaxPooling2D((2, 2)),        layers.Conv2D(64, (3, 3), activation='relu'),        layers.Flatten(),        layers.Dense(64, activation='relu'),        layers.Dense(10, activation='softmax')    ])
    model.compile(optimizer='adam',                  loss='sparse_categorical_crossentropy',                  metrics=['accuracy'])
    # 训练模型    print("开始训练模型...")    history = model.fit(train_images, train_labels, epochs=5, batch_size=32,              validation_split=0.2, verbose=1)
    # 评估模型    test_loss, test_acc = model.evaluate(test_images, test_labels, verbose=2)    print(f"测试准确率: {test_acc:.4f}")
    return model, history, test_acc

New-ItemProperty -Path "HKLM:\SYSTEM\CurrentControlSet\Control\FileSystem" `-Name "LongPathsEnabled" -Value 1 -PropertyType DWORD -Force

static ai_handle network = AI_HANDLE_NULL;
/* Global c-array to handle the activations buffer *//*ڴȫc*/AI_ALIGNED(32)static ai_u8 activations[AI_NETWORK_DATA_ACTIVATIONS_SIZE];
/* Array to store the data of the input tensor *//*ڴ洢?*/AI_ALIGNED(32)static ai_float in_data[AI_NETWORK_IN_1_SIZE]; // 这里假设 AI_NETWORK_IN_1_SIZE 为 1600 (40x40)/* or static ai_u8 in_data[AI_NETWORK_IN_1_SIZE_BYTES]; *//*ڴ洢ݵc*/
/* c-array to store the data of the output tensor */AI_ALIGNED(32)static ai_float out_data[AI_NETWORK_OUT_1_SIZE];/* static ai_u8 out_data[AI_NETWORK_OUT_1_SIZE_BYTES]; */
/* Array of pointer to manage the model's input/output tensors *//*ڹģ/ָ*/static ai_buffer *ai_input;static ai_buffer *ai_output;
/*  * Bootstrap */int aiInit(void) {  ai_error err;    /* Create and initialize the c-model *//*ʼc-model */  const ai_handle acts[] = { activations };  err = ai_network_create_and_init(&network, acts, NULL);  if (err.type != AI_ERROR_NONE) {  };
  /* Reteive pointers to the model's input/output tensors *//*ȡָģ/ָ*/  ai_input = ai_network_inputs_get(network, NULL);  ai_output = ai_network_outputs_get(network, NULL);
  return 0;}
/*  * Run inference */int aiRun(const void *in_data, void *out_data) {  ai_i32 n_batch;  ai_error err;    /* 1 - Update IO handlers with the data payload *//* 1 -ЧغɸIO*/  ai_input[0].data = AI_HANDLE_PTR(in_data);  ai_output[0].data = AI_HANDLE_PTR(out_data);
  /* 2 - Perform the inference *//* 2 -ִ*/  n_batch = ai_network_run(network, &ai_input[0], &ai_output[0]);  if (n_batch != 1) {    err = ai_network_get_error(network);    HAL_GPIO_TogglePin(GPIOF,GPIO_PIN_10);  };    return 0;}

  while(1)  {    tp_dev.scan(1);    if(tp_dev.sta == 0xC1)    {      uint16_t x = tp_dev.x[0];      uint16_t y = tp_dev.y[0];            // 计算触摸点所在的网格位置      int gridX = x / 10;      int gridY = y / 10;            // 标记该网格为已触摸，并点亮对应的10x10方块      if(gridX < 40 && gridY < 40) {        int index = gridY * 40 + gridX;        if(in_data[index] == 0) {          in_data[index] = 1.0f;          LCD_Fill(gridX*10, gridY*10, (gridX+1)*10-1, (gridY+1)*10-1, RED);        }      }    }  }

KeyCallBack(&key,GPIOE,GPIO_PIN_2);if(key.HasPressNum){  HAL_GPIO_TogglePin(GPIOF,GPIO_PIN_9);  key.HasPressNum = 0;  aiRun(in_data,out_data);  // 找出 out_data 中的最大值及其索引  int max_index = 0;  float max_value = out_data[0];  for (int i = 1; i < AI_NETWORK_OUT_1_SIZE; i++) {      if (out_data[i] > max_value) {          max_value = out_data[i];          max_index = i;        }   }          // 在这里可以使用 max_index 和 max_value，将结果显示在LCD上或通过其他方式输出          char result[20];  sprintf(result, "The most like: %d (%.2f)", max_index, max_value);  LCD_ShowString(10, 410, 400, 32, 32, (uint8_t*)result);  clearAllTouchedBlocks(); // 清除所有点亮的方块  }