# 비디오의 프레임에 Semantic segmentation 적용

모델에 파이프라인 명령어를 실행하기 전에, 필요한 [전제 조건](https://docs.qualcomm.com/doc/80-70018-15BK/topic/ai-hub-qualcomm-im-sdk.html#prerequisites) 을 따라야 합니다.

다음 명령어를 실행하여 결과가 연결된 디스플레이에 표시되는지 확인하세요.

export XDG_RUNTIME_DIR=/dev/socket/weston && export WAYLAND_DISPLAY=wayland-1
    Copy to clipboard

## DeepLabV3-Plus\_MobileNet-Quantized

DeepLabV3 Quantized는 여러 규모에서 Semantic segmentation 위해 설계되었으며 다양한 데이터 세트에서 학습되었습니다.

AI Hub 모델은 [이 DeepLabV3-Plus-MobileNet-Quantized 구현](https://github.com/jfzhang95/pytorch-deeplab-xception) 을 기반으로 합니다.

- 모델: [deeplabv3_plus_mobilenet_quantized.tflite](https://aihub.qualcomm.com/iot/models/deeplabv3_plus_mobilenet_quantized)
- 레이블: [voc_labels.txt](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/labels/voc_labels.txt)

gst-launch-1.0 -e --gst-debug=2 \
    filesrc location=/etc/media/video.mp4 ! qtdemux ! queue ! h264parse ! v4l2h264dec capture-io-mode=4 output-io-mode=4 ! video/x-raw,format=NV12 ! 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=/etc/models/deeplabv3_plus_mobilenet_quantized.tflite ! queue ! \
    qtimlvsegmentation module=deeplab-argmax labels=/etc/labels/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

FCN\_ResNet50은 COCO 데이터 세트에서 이미지를 분류할 수 있는 양자화된 머신 러닝 모델입니다.

AI Hub 모델은 [이 FCN-ResNet50-Quantized 구현](https://github.com/pytorch/vision/blob/main/torchvision/models/segmentation/fcn.py) 을 기반으로 합니다.

- 모델: [FCN-ResNet50-Quantized.tflite](https://aihub.qualcomm.com/iot/models/fcn_resnet50_quantized)
- 레이블: [voc_labels.txt](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/labels/voc_labels.txt)

참고

이 파이프라인은 현재 QCS6490에서 지원되지 않습니다.

gst-launch-1.0 -e --gst-debug=2 \
    filesrc location=/etc/media/video.mp4 ! qtdemux ! queue ! h264parse ! v4l2h264dec capture-io-mode=4 output-io-mode=4 ! video/x-raw,format=NV12 ! 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=/etc/models/fcn_resnet50_quantized.tflite ! queue ! \
    qtimlvsegmentation module=deeplab-argmax labels=/etc/labels/voc_labels.txt \
    constants="deeplab,q-offsets=<0.0>,q-scales=<1.0>;" ! video/x-raw,format=BGRA,width=256,height=144 ! queue ! mixer.
    Copy to clipboard

## FFNet-40S-Quantized

FFNet-40S-Quantized는 도로, 보도, 보행자 등 픽셀별 클래스로 거리 장면 이미지를 구분하는 “fuss-free network”입니다.

cityscapes 데이터 세트에서 학습되었습니다.

AI Hub 모델은 [이 FFNet-40S-Quantized 구현](https://github.com/Qualcomm-AI-research/FFNet) 을 기반으로 합니다.

- 모델: [ffnet_40s_quantized.tflite](https://aihub.qualcomm.com/iot/models/ffnet_40s_quantized)
- 레이블: [voc_labels.txt](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/labels/voc_labels.txt)

gst-launch-1.0 -e --gst-debug=2 \
    filesrc location=/etc/media/video.mp4 ! qtdemux ! queue ! h264parse ! v4l2h264dec capture-io-mode=4 output-io-mode=4 ! video/x-raw,format=NV12 ! queue ! tee name=split \
    split. ! queue ! qtivcomposer name=mixer sink_1::alpha=0.5 ! queue ! waylandsink sync=false fullscreen=true \
    split. ! queue ! qtimlvconverter ! queue ! qtimltflite delegate=external external-delegate-path=libQnnTFLiteDelegate.so \
    external-delegate-options="QNNExternalDelegate,backend_type=htp;" model=/etc/models/ffnet_40s_quantized.tflite ! queue ! \
    qtimlvsegmentation module=deeplab-argmax labels=/etc/labels/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

FFNet-54S-Quantized는 도로, 보도, 보행자 등 픽셀별 클래스로 거리 장면 이미지를 구분하는 “fuss-free network”입니다.

cityscapes 데이터 세트에서 학습되었습니다.

AI Hub 모델은 [이 FFNet-54S-Quantized 구현](https://github.com/Qualcomm-AI-research/FFNet) 을 기반으로 합니다.

- 모델: [ffnet_54s_quantized.tflite](https://aihub.qualcomm.com/iot/models/ffnet_54s_quantized)
- 레이블: [voc_labels.txt](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/labels/voc_labels.txt)

gst-launch-1.0 -e --gst-debug=2 \
    filesrc location=/etc/media/video.mp4 ! qtdemux ! queue ! h264parse ! v4l2h264dec capture-io-mode=4 output-io-mode=4 ! video/x-raw,format=NV12 ! queue ! tee name=split \
    split. ! queue ! qtivcomposer name=mixer sink_1::alpha=0.5 ! queue ! waylandsink sync=false fullscreen=true \
    split. ! queue ! qtimlvconverter ! queue ! qtimltflite delegate=external external-delegate-path=libQnnTFLiteDelegate.so \
    external-delegate-options="QNNExternalDelegate,backend_type=htp;" model=/etc/models/ffnet_54s_quantized.tflite ! queue ! \
    qtimlvsegmentation module=deeplab-argmax labels=/etc/labels/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

FFNet-78S-Quantized는 도로, 보도, 보행자 등 픽셀별 클래스로 거리 장면 이미지를 구분하는 “fuss-free network”입니다.

cityscapes 데이터 세트에서 학습되었습니다.

AI Hub 모델은 [이 FFNet-78S-Quantized 구현](https://github.com/Qualcomm-AI-research/FFNet) 을 기반으로 합니다.

- 모델: [ffnet_78s_quantized.tflite](https://aihub.qualcomm.com/iot/models/ffnet_78s_quantized)
- 레이블: [voc_labels.txt](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/labels/voc_labels.txt)

gst-launch-1.0 -e --gst-debug=2 \
    filesrc location=/etc/media/video.mp4 ! qtdemux ! queue ! h264parse ! v4l2h264dec capture-io-mode=4 output-io-mode=4 ! video/x-raw,format=NV12 ! queue ! tee name=split \
    split. ! queue ! qtivcomposer name=mixer sink_1::alpha=0.5 ! queue ! waylandsink sync=false fullscreen=true \
    split. ! queue ! qtimlvconverter ! queue ! qtimltflite delegate=external external-delegate-path=libQnnTFLiteDelegate.so \
    external-delegate-options="QNNExternalDelegate,backend_type=htp;" model=/etc/models/ffnet_78s_quantized.tflite ! queue ! \
    qtimlvsegmentation module=deeplab-argmax labels=/etc/labels/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: Oct 15, 2025

[Previous Topic
객체 감지](https://docs.qualcomm.com/bundle/publicresource/80-70018-15BK/topics/object-detection.md) [Next Topic
초고해상도로 이미지 업스케일링](https://docs.qualcomm.com/bundle/publicresource/80-70018-15BK/topics/super-resolution.md)