NVIDIA VideoFX SDK使用

环境搭建

1. docker pull qiushenjie/cuda11.1-cudnn8-devel-ubuntu18.04-python3.6:latest
2. download tensorrt https://developer.nvidia.com/compute/machine-learning/tensorrt/secure/7.2.2/tars/TensorRT-7.2.2.3.CentOS-7.6.x86_64-gnu.cuda-11.1.cudnn8.0.tar.gz
3. download NVIDIA Video Effects SDK https://developer.nvidia.com/maxine-getting-started

---

 

SDK要求的环境必须的软件版本

没办法，笔者只能在docker容器中来操作了

cmake >= 3.10.x

Ubuntu==18.04

TensorRT==7.2.2.3

CUDA==11.1

cuDNN==8.x

---

 

解压到容器的/usr/local/下

tar -zxvf TensorRT-7.2.2.3.Ubuntu-18.04.x86_64-gnu.cuda-11.1.cudnn8.0.tar.gz -C /usr/local
tar -zxvf VideoFX-ubuntu18.04-x86_64-0.6.0.0.tar.gz -C /usr/local

编译SDK的示例

cd VideoFX/share

./build_samples.sh

[yes]

\n

最终结果可以在[yes]这部输入指定路径，否则示例默认安装到~/mysamples

运行示例

cd ~/mysamples

. ./setup_env.sh

VideoEffectsApp \
        --model_dir=$_VFX_MODELS \
        --in_file=$_VFX_SHARE/samples/input/input2.jpg \
        --out_file=sr_1.png \
        --effect=SuperRes \
        --resolution=2160 \
        --strength=1 \
        --show=false \
    　　--verbose=true

参数说明请参考https://developer.nvidia.com/sites/default/files/akamai/maxine/NVIDIA_Video_Effects_SDK_Programming_Guide_Linux.pdf