# 在應用程序中集成AI Hub模型

本節描述如何在參考應用程序中使用AI Hub模型。此示例使用來自AI Hub的圖像分類模型。

備註

大多數公開可用的圖像分類模型的預處理和後處理是相似的，並且可能由 `qtimlvclassification` 外掛程式支持。

Qualcomm AI Hub為Qualcomm設備提供優化的AI模型，允許它們使用LiteRT（以前稱為Tensorflow Lite）或Qualcomm® AI Engine Direct在CPU、GPU或NPU上運行。

此示例解釋了如何使用來自AI Hub的 [mobilenet-v2-quantized](https://aihub.qualcomm.com/iot/models/mobilenet_v2_quantized) 模型並將該模型集成到 `gst-ai-classification` 參考應用程序中。

| **序號** | **參考應用** | **已驗證的AI Hub模型** | **標籤檔案** |
| --- | --- | --- | --- |
| 1 | gst-ai-classification | googlenet\_quantized | [標籤](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/labels/imagenet_labels.txt) |
| 2 | gst-ai-classification | inception\_v3\_quantized | [標籤](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/labels/imagenet_labels.txt) |
| 3 | gst-ai-classification | mobilenet-v2-quantized | [標籤](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/labels/imagenet_labels.txt) |
| 4 | gst-ai-classification | MobileNet-v3-Large-Quantized | [標籤](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/labels/imagenet_labels.txt) |
| 5 | gst-ai-classification | ResNet101Quantized | [標籤](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/labels/imagenet_labels.txt) |
| 6 | gst-ai-classification | SqueezeNet-1\_1Quantized | [標籤](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/labels/imagenet_labels.txt) |
| 7 | gst-ai-classification | ResNet18Quantized | [標籤](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/labels/imagenet_labels.txt) |
| 8 | gst-ai-classification | resnext50\_quantized | [標籤](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/labels/imagenet_labels.txt) |
| 9 | gst-ai-classification | wideresnet50\_quantized | [標籤](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/labels/imagenet_labels.txt) |
| 10 | gst-ai-classification | shufflenet\_v2\_quantized | [標籤](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/labels/imagenet_labels.txt) |
| 11 | gst-ai-classification | resnext101\_quantized | [標籤](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/labels/imagenet_labels.txt) |
| 12 | gst-ai-segmentation | deeplabv3\_plus\_mobilenet\_quantized | [標籤](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/labels/voc_labels.txt) |
| 13 | gst-ai-segmentation | fcn\_resnet50\_quantized | [標籤](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/labels/voc_labels.txt) |
| 14 | gst-ai-segmentation | ffnet\_40s\_quantized | [標籤](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/labels/voc_labels.txt) |
| 15 | gst-ai-segmentation | ffnet\_54s\_quantized | [標籤](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/labels/voc_labels.txt) |
| 16 | gst-ai-segmentation | ffnet\_78s\_quantized | [標籤](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/labels/voc_labels.txt) |
| 17 | gst-ai-object-detection | Yolo-NAS-Quantized | [Labels](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/labels/coco_labels.txt) |
| 18 | gst-ai-object-detection | Yolo-v7-Quantized | [Labels](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/labels/coco_labels.txt) |
| 19 | gst-ai-object-detection | YOLOv8-Detection-Quantized | [Labels](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/labels/coco_labels.txt) |
| 20 | gst-ai-superresolution | QuickSRNetLarge-Quantized | 不適用 |
| 21 | gst-ai-superresolution | QuickSRNetMedium-Quantized | 不適用 |
| 22 | gst-ai-superresolution | QuickSRNetSmall-Quantized | 不適用 |
| 23 | gst-ai-superresolution | XLSR-Quantized | 不適用 |
|  |  |  |  |
|  |  |  |  |
|  |  |  |  |
|  |  |  |  |
|  |  |  |  |
|  |  |  |  |
|  |  |  |  |
|  |  |  |  |
|  |  |  |  |
|  |  |  |  |

備註

一些模型只有LiteRT或Qualcomm AI Engine Direct SDK模型可在AI Hub中使用。然而，AI Hub經常更新其庫以包含更多模型並增強現有模型。

先決條件

1. 下載 [AI Hub模型](https://aihub.qualcomm.com/iot/models)。
2. 選擇要下載模型的晶片組。

    - 對於RB3 Gen 2，選擇 *Qualcomm QCS6490*
    - 對於QCS9075，選擇 *Qualcomm QCS9075*
    - 對於QCS8275，選擇 *Qualcomm QCS8275*
3. 將與模型匹配的標籤檔案複製到設備上。

    標籤檔案可在提供的 [鏈接](https://github.com/quic/ai-hub-models/tree/main/qai_hub_models/labels) 下載。

## 獲取模型常數

1. 使用圖形查看工 如 [Netron](https://netron.app/) 打開下載的LiteRT模型。
2. 檢查要包含在參考應用程序中的特定模型的後處理要求。

    1. 選擇模型的輸入節點以查看模型屬性。
    2. 從模型的輸出節點複製值。

![../_images/execute-mobilenet-v2-model-properties.png](data:image/png;base64,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)

    例如，在上圖中，`q_scale` 是量化：線性中提到的值，並且 `q_offset` 在輸出時為 `-59` 。您需要更改模型圖中的 `q_offset` 符號，這是根據外掛程式的實現所需的。

記下這些值，並在修改配置時使用它們。

## 修改配置

1. 將模型複製到設備。

scp mobilenet_v2_quantized.tflite root@<IP address of the target device>:/etc/models/
        Copy to clipboard
2. 將標籤複製到設備。

wget https://raw.githubusercontent.com/quic/ai-hub-models/refs/heads/main/qai_hub_models/labels/imagenet_labels.txt
        Copy to clipboard

scp imagenet_labels.txt root@<IP addr of the target device>:/etc/labels/
        Copy to clipboard
3. 如下所示更新 `/etc/configs/config_classification.json` 檔案。

{
        "file-path": "/etc/media/video.mp4",
        "ml-framework": "tflite",
        "model": "/etc/models/mobilenet_v2_quantized.tflite",
        "labels": "/etc/labels/imagenet_labels.txt",
        "constants": "Mobilenet,q-offsets=<63.0>,q-scales=<0.16931529343128204>;",
        "threshold": 40,
        "runtime": "dsp"
        }
        Copy to clipboard

備註

使用在獲 [取模型常數](https://docs.qualcomm.com/doc/80-70018-15BT/topic/integrate-ai-hub-models.html#get-model-constants) 中記錄的值更新 `q-offset` 和 `q_scale`  。

## 運行示例應用

1. 通過運行以下命令啟用顯示的 Weston 服務：

export XDG_RUNTIME_DIR=/dev/socket/weston && export WAYLAND_DISPLAY=wayland-1
        Copy to clipboard
2. 通過在命令行參數中給出常數值來運行參考應用程序：

gst-ai-classification --config-file=/etc/configs/config_classification.json
        Copy to clipboard
3. 使用示例應用程序在您的設備上運行各種AI Hub模型。要運行不同的AI Hub模型：

    1. 下載您要運行的模型和標籤檔案。
    2. 將模型和標籤檔案複製到設備上。
    3. 使用 [Netron](https://netron.app/) 應用程序檢查模型檔案，並相應地填寫常數參數。
    4. 按照下載模型檔案 download models files 中的步驟下載YOLOv8量化LiteRT模型檔案（用於對象偵測）。
    5. 如下所示更新 `/etc/configs/config_detection.json` 檔案。

{
            "file-path": "/etc/media/video.mp4",
            "ml-framework": "tflite",
            "yolo-model-type": "yolov8",
            "model": "/etc/models/YOLOv8-Detection-Quantized.tflite",
            "labels": "/etc/labels/yolonas.labels",
            "constants": "YOLOv8,q-offsets=<21.0, 0.0, 0.0>,q-scales=<3.093529462814331, 0.00390625, 1.0>;",
            "threshold": 40,
            "runtime": "dsp"
            }
            Copy to clipboard

備註

根據獲取的模型更新配置檔案中的模型名稱和模型常數。
    6. 運行示例應用程序。

備註

確保已將模型和標籤檔案複製到設備上。

gst-ai-object-detection --config-file=/etc/configs/config_detection.json
            Copy to clipboard
    7. 如下所示更新 `/etc/configs/config_detection.json` 檔案：

        以 `deeplabv3_plus_mobilenet_quantized.tflite` 模型檔案（用於分割）和 `deeplabv3_resnet50.labels` 檔案為例。

{
            "file-path": "/etc/media/video.mp4",
            "ml-framework": "tflite",
            "model": "/etc/models/deeplabv3_plus_mobilenet_quantized.tflite",
            "labels": "/etc/labels/deeplabv3_resnet50.labels",
            "constants": "deeplab,q-offsets=<0.0>,q-scales=<1.0>;",
            "runtime": "dsp"
            }
            Copy to clipboard
    8. 運行示例應用程序。

gst-ai-segmentation --config-file=/etc/configs/config_segmentation.json
            Copy to clipboard
    9. 以 `quicksrnetsmall_quantized.tflite` 模型檔案（用於超分辨率）為例運行：

gst-ai-superresolution --input-file=/etc/media/video.mp4 --model=/etc/models/quicksrnetsmall_quantized.tflite --constants="srnet,q-offsets=<0.0>,q-scales=<1.0>;"
            Copy to clipboard

備註

確保使用的影片具有128x128的分辨率，這是模型的輸入。
    10. 有關輸入參數的更多詳細信息，請參閱應用程序的幫助部分。

備註

在分割情況下，未識別的對象可能會出現在白色背景上。此問題源於標籤檔案，將在未來的更新中解決。
4. 如果要進行任何更改並運行它，請重新編譯參考應用程序。請參閱 [Qualcomm智慧多媒體軟體開發工具包（IM SDK）快速入門指南](https://docs.qualcomm.com/bundle/publicresource/topics/80-70018-51/introduction.html) ，了解如何使用eSDK編譯參考應用程序的說明。

Last Published: Oct 15, 2025

[Previous Topic
在應用程序中集成自定義AI模型](https://docs.qualcomm.com/bundle/publicresource/80-70018-15BT/topics/integrate-custom-model.md) [Next Topic
添加自訂模型的後處理支援](https://docs.qualcomm.com/bundle/publicresource/80-70018-15BT/topics/add-postprocessing-support-custom-model.md)

Source: [https://docs.qualcomm.com/doc/80-70018-15BT/topic/integrate-ai-hub-models.html](https://docs.qualcomm.com/doc/80-70018-15BT/topic/integrate-ai-hub-models.html)