# 语义分割

Source: [https://docs.qualcomm.com/doc/80-70015-15BY/topic/semantic-segmentation.html](https://docs.qualcomm.com/doc/80-70015-15BY/topic/semantic-segmentation.html)

在对模型执行 pipeline 命令之前，请确保遵循所需的 [先决条件](https://docs.qualcomm.com/doc/80-70015-15BY/topic/ai-hub-imsdk.html#ai_hub_imsdk__aihub-imsdk-prereq)

## DeepLabV3-Plus\_MobileNet-Quantized

    gst-launch-1.0 -e --gst-debug=2 \
    filesrc location=/opt/video.mp4 ! qtdemux ! queue ! h264parse ! v4l2h264dec capture-io-mode=5 output-io-mode=5 ! queue ! tee name=split \
    split. ! queue ! qtivcomposer name=mixer sink_1::alpha=0.5 ! queue ! waylandsink sync=true fullscreen=true \
    split. ! queue ! qtimlvconverter ! queue ! qtimltflite delegate=external external-delegate-path=libQnnTFLiteDelegate.so \
    external-delegate-options="QNNExternalDelegate,backend_type=htp;" model=/opt/deeplabv3_plus_mobilenet_quantized.tflite ! queue ! \
    qtimlvsegmentation module=deeplab-argmax labels=/opt/voc_labels.txt \
    constants="deeplab,q-offsets=<92.0>,q-scales=<0.04518842324614525>;" ! video/x-raw,format=BGRA,width=256,height=144 ! queue ! mixer.Copy to clipboard

## FCN-Resnet50-Quantized

Note: QCS6490 目前不支持此 pipeline。

    gst-launch-1.0 -e --gst-debug=2 \
    filesrc location=/opt/video.mp4 ! qtdemux ! queue ! h264parse ! v4l2h264dec capture-io-mode=5 output-io-mode=5 ! queue ! tee name=split \
    split. ! queue ! qtivcomposer name=mixer sink_1::alpha=0.5 ! queue ! waylandsink sync=true fullscreen=true \
    split. ! queue ! qtimlvconverter ! queue ! qtimltflite delegate=external external-delegate-path=libQnnTFLiteDelegate.so \
    external-delegate-options="QNNExternalDelegate,backend_type=htp;" model=/opt/fcn_resnet50_quantized.tflite ! queue ! \
    qtimlvsegmentation module=deeplab-argmax labels=/opt/voc_labels.txt constants="fcnresnet,q-offsets=<70.0>,q-scales=<0.08319790661334991>;" ! \
    video/x-raw,format=BGRA,width=256,height=144 ! queue ! mixer.Copy to clipboard

## FFNet-40S-Quantized

    gst-launch-1.0 -e --gst-debug=2 \
    filesrc location=/opt/video.mp4 ! qtdemux ! queue ! h264parse ! v4l2h264dec capture-io-mode=5 output-io-mode=5 ! queue ! tee name=split \
    split. ! queue ! qtivcomposer name=mixer sink_1::alpha=0.5 ! queue ! waylandsink sync=true fullscreen=true \
    split. ! queue ! qtimlvconverter ! queue ! qtimltflite delegate=external external-delegate-path=libQnnTFLiteDelegate.so \
    external-delegate-options="QNNExternalDelegate,backend_type=htp;" model=/opt/ffnet_40s_quantized.tflite ! queue ! \
    qtimlvsegmentation module=deeplab-argmax labels=/opt/voc_labels.txt constants="ffnet,q-offsets=<178.0>,q-scales=<0.31378185749053955>;" ! \
    video/x-raw,format=BGRA,width=256,height=144 ! queue ! mixer.Copy to clipboard

## FFNet-54S-Quantized

    gst-launch-1.0 -e --gst-debug=2 \
    filesrc location=/opt/video.mp4 ! qtdemux ! queue ! h264parse ! v4l2h264dec capture-io-mode=5 output-io-mode=5 ! queue ! tee name=split \
    split. ! queue ! qtivcomposer name=mixer sink_1::alpha=0.5 ! queue ! waylandsink sync=true fullscreen=true \
    split. ! queue ! qtimlvconverter ! queue ! qtimltflite delegate=external external-delegate-path=libQnnTFLiteDelegate.so \
    external-delegate-options="QNNExternalDelegate,backend_type=htp;" model=/opt/ffnet_54s_quantized.tflite ! queue ! \
    qtimlvsegmentation module=deeplab-argmax labels=/opt/voc_labels.txt constants="ffnet,q-offsets=<178.0>,q-scales=<0.2929433584213257>;" ! \
    video/x-raw,format=BGRA,width=256,height=144 ! queue ! mixer.Copy to clipboard

## FFNet-78S-Quantized

    gst-launch-1.0 -e --gst-debug=2 \
    filesrc location=/opt/video.mp4 ! qtdemux ! queue ! h264parse ! v4l2h264dec capture-io-mode=5 output-io-mode=5 ! queue ! tee name=split \
    split. ! queue ! qtivcomposer name=mixer sink_1::alpha=0.5 ! queue ! waylandsink sync=true fullscreen=true \
    split. ! queue ! qtimlvconverter ! queue ! qtimltflite delegate=external external-delegate-path=libQnnTFLiteDelegate.so \
    external-delegate-options="QNNExternalDelegate,backend_type=htp;" model=/opt/ffnet_78s_quantized.tflite ! queue ! \
    qtimlvsegmentation module=deeplab-argmax labels=/opt/voc_labels.txt constants="ffnet,q-offsets=<171.0>,q-scales=<0.3849360942840576>;" ! \
    video/x-raw,format=BGRA,width=256,height=144 ! queue ! mixer.Copy to clipboard

Last Published: Jan 26, 2026

[Previous Topic
目标检测](https://docs.qualcomm.com/bundle/publicresource/80-70015-15BY/topics/object-detection.md) [Next Topic
超分辨率](https://docs.qualcomm.com/bundle/publicresource/80-70015-15BY/topics/super-resolution.md)