# 设置 AI Hub 以优化 AI 模型

为了在 Qualcomm AI 硬件上快速构建模型原型，AI Hub 提供了一种方法，帮助开发者针对机器视觉、音频和语音用例在设备上对机器学习模型进行优化、验证和部署

![../_images/ai-hub_QLI.png](data:image/png;base64,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)

## 设置您的环境

1. 设置 Python 环境。在计算机上安装 [miniconda](https://docs.conda.io/projects/miniconda/en/latest/miniconda-install.html)。

Tab Windows
Tab macOS/Linux

安装完成后，从“开始”菜单打开 Anaconda 提示符。

安装完成后，打开一个新的 shell 窗口。

    为 AI Hub 搭建 Python 虚拟环境。

conda activate
        Copy to clipboard

conda create python=3.10 -n qai_hub
        Copy to clipboard

conda activate qai_hub
        Copy to clipboard
2. 安装 git.

sudo apt-get install git
        Copy to clipboard
3. 安装 AI Hub Python 客户端。

pip3 install qai-hub
        Copy to clipboard

pip3 install "qai-hub[torch]"
        Copy to clipboard
4. 登录 AI Hub。

    前往 [AI Hub](https://aihub.qualcomm.com/) 并使用 Qualcomm ID 登录，查看所创建的作业的相关信息。

    登录后，转至 *Account &gt; Settings &gt; API Token*。此时应提供一个可用于配置客户端的 API 令牌。
5. 在终端，使用以下命令通过 API 令牌配置客户端。

qai-hub configure --api_token <INSERT_API_TOKEN>
        Copy to clipboard

## 选择 AI Hub 工作流程

### 使用预优化模型

1. 转到 [AI Hub Model Zoo](https://aihub.qualcomm.com/iot/models) 以访问可用于 Qualcomm 评估套件的预优化模型。
2. 筛选可用于您的 EVK 的模型。例如，通过在左侧窗格中选择 *Qualcomm QCS6490* 作为芯片组，可以下载针对 Qualcomm Dragonwing™ RB3 Gen 2 的预优化模型。
3. 从筛选视图中选择一个模型以转到模型页面。
4. 在模型页面上，选择 runtime 和精度。
5. 选择下载以开始模型下载。下载的模型已经过预优化，可直接部署。如需了解更多部署模型的详细信息，请参阅 [开发您自己的 AI/ML 应用程序](https://docs.qualcomm.com/doc/80-70022-15BY/topic/develop-your-own-application.html)。

![../_images/ai-hub-download.png](data:image/png;base64,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)

### 引入您自己的模型

1. 选择 PyTorch 或 Onnx 格式的预训练模型。
2. 使用 Python API 将模型提交至 AI Hub 以进行编译或优化。

    提交编译作业时，您必须为 EVK 选择设备或芯片组以及编译模型的目标 runtime。对于 RB3Gen2，支持 LiteRT runtime。

    | **芯片组** | **Runtime** | **CPU** | **GPU** | **HTP** |
    | --- | --- | --- | --- | --- |
    | Qualcomm Dragonwing™ RB3 Gen 2 | LiteRT | INT8,FP16, FP32 | FP16,FP32 | INT8, INT16 |

    提交后，AI Hub 会为该作业生成一个唯一的 ID。用户可以使用此作业 ID 查看作业详情。
3. AI Hub 会根据选择的设备和 runtime 对模型进行优化。

    - 或者，也可以提交作业以便在源自设备集群且已经过配置的实际设备上对优化模型进行分析或推理（使用 Python API）。

        - Profiling：在已配置的设备上对模型进行基准测试并提供统计数据，包括层级的平均推理时间、runtime 配置等。
        - 推理：作为推理作业执行过程中的一部分，基于提交的数据使用优化模型在已配置的设备上运行该模型进行推理通过。
4. 提交的每项作业都可以在 AI Hub 门户中进行审核。提交编译作业时，将提供优化模型的可下载链接。然后，该优化模型可以部署在 RB3Gen2 等本地开发设备上。

下方所示为从 [AI Hub documentation](https://app.aihub.qualcomm.com/docs/) 中获取的所述工作流程的示例。在本示例中，会将源自 PyTorch 的 MobileNet V2 预训练模型上传到 AI Hub，并编译为优化的 TFLite 模型以便在 RB3Gen2 芯片组上运行。

import qai_hub as hub
    import torch
    from torchvision.models import mobilenet_v2
    import numpy as np
    
    # Using pre-trained MobileNet
    torch_model = mobilenet_v2(pretrained=True)
    torch_model.eval()
    
    # Trace model (for on-device deployment)
    input_shape = (1, 3, 224, 224)
    example_input = torch.rand(input_shape)
    traced_torch_model = torch.jit.trace(torch_model, example_input)
    
    # Compile and optimize the model for a specific device
    compile_job = hub.submit_compile_job(
        model=traced_torch_model,
        device=hub.Device("QCS6490 (Proxy)"),
        input_specs=dict(image=input_shape),
        #compile_options="--target_runtime tflite",
    )
    
    # Profiling Job
    profile_job = hub.submit_profile_job(
        model=compile_job.get_target_model(),
        device=hub.Device("QCS6490 (Proxy)"),
    )
    
    sample = np.random.random((1, 3, 224, 224)).astype(np.float32)
    
    # Inference Job
    inference_job = hub.submit_inference_job(
        model=compile_job.get_target_model(),
        device=hub.Device("QCS6490 (Proxy)"),
        inputs=dict(image=[sample]),
    )
    
    # Download model
    compile_job.download_target_model(filename="/tmp/mobilenetv2.tflite")
    Copy to clipboard

Note

请使用以下命令，以便停用先前激活的 `qai_hub` 环境。

conda deactivate
    Copy to clipboard

下载模型后，即可使用它来 [开发您自己的 AI/ML 应用程序](https://docs.qualcomm.com/doc/80-70022-15BY/topic/develop-your-own-application.html)。

如需了解更多 AI Hub 工作流程和 API 的详细信息，请参阅 [AI Hub 文档](https://app.aihub.qualcomm.com/docs/hub/index.html#examples)，浏览 [AI Hub 教程视频](https://www.youtube.com/watch?v=V1CDWYZ7Shw&amp;list=PLxeazpXYyqtOowtUdvigvAgMV5_K1KIrh)，或观看以下有关如何在 AI Hub 中 profile 模型的视频。

<svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" viewbox="0 0 800 500" width="800" height="500" style="cursor:auto !important" aria-label="../_images/ai-hub-video.svg" svgdefaultwidth="75%">
    <defs>
      <style>@import url("https://fonts.googleapis.com/css2?family=Roboto+Flex:opsz,wght@8..144,100..1000&amp;display=swap");
.svg-1 .bg-fill { fill: var(--color-background) }
.svg-1 .fill-text { color: var(--color-content); fill: var(--color-content) }
.svg-1 .video-hoverbox { transition: opacity 0.15s ease-in-out }
.svg-1 .video-hoverbox:hover { opacity: 0.9 }</style>
  </defs>

  <foreignobject x="0" y="0" width="800" height="500">
    <body xmlns="http://www.w3.org/1999/xhtml">
        <iframe width="800" height="500" src="https://players.brightcove.net/1414329538001/4JiZQnWhg_default/index.html?videoId=6366848482112" allowfullscreen="" allow="encrypted-media"></iframe>
    <div class='topic-detail'><div class='topic-updated-date'><span> Last Published: </span>Nov 05, 2025</div><div class='prev-and-next-links'><span class='previous-topic-link'><span aria-hidden='true' class='disabled' data-tip='' data-effect='solid'></span></span></div></div></body>
    </foreignobject>
</svg>

Note

以上视频以 Python 3.8 为例。

支持 Python 3.8 和 Python 3.10。

Last Published: Nov 05, 2025

[Previous Topic
编译和优化 AI 模型](https://docs.qualcomm.com/bundle/publicresource/80-70022-15BY/topics/compile-and-optimize-model.md) [Next Topic
使用 LiteRT 优化 AI 模型](https://docs.qualcomm.com/bundle/publicresource/80-70022-15BY/topics/tflite.md)