私的AI研究会 > FaceFusion
1枚の顔画像から動画内の顔を簡単に置き換えることのできる FaceFusion を検証する
(base) nvidia-smi Tue Nov 19 05:22:45 2024 +-----------------------------------------------------------------------------------------+ | NVIDIA-SMI 555.99 Driver Version: 555.99 CUDA Version: 12.5 | |-----------------------------------------+------------------------+----------------------+ | GPU Name Driver-Model | Bus-Id Disp.A | Volatile Uncorr. ECC | | Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. | | | | MIG M. | |=========================================+========================+======================| | 0 NVIDIA GeForce RTX 4070 Ti WDDM | 00000000:01:00.0 On | N/A | | 31% 26C P8 4W / 285W | 1275MiB / 12282MiB | 1% Default | | | | N/A | +-----------------------------------------+------------------------+----------------------+ :
(base) nvcc -V nvcc: NVIDIA (R) Cuda compiler driver Copyright (c) 2005-2023 NVIDIA Corporation Built on Fri_Nov__3_17:51:05_Pacific_Daylight_Time_2023 Cuda compilation tools, release 12.3, V12.3.103 Build cuda_12.3.r12.3/compiler.33492891_0
(facefusion) conda deactivate (base) conda env remove -n facefusion
(base) conda create -n facefusion python=3.12 -y (base)conda info -e
(base) conda activate facefusion
(facefusion) conda install conda-forge::cuda-runtime=12.4.1 conda-forge::cudnn=9.2.1.18 -y (facefusion) conda install ffmpeg ffmpeg-python -c conda-forge -y (facefusion) pip install tensorrt==10.5.0 --extra-index-url https://pypi.nvidia.com
(facefusion) cd workspace_3 (facefusion) pip install -r req_facefusion.txt
(facefusion) git clone https://github.com/facefusion/facefusion (facefusion) cd facefusion (facefusion) python install.py --onnxruntime default
(facefusion) pip uninstall onnxruntime (facefusion) pip install onnxruntime-gpu
(facefusion) python facefusion.py run
[FACEFUSION.CORE] Processing step 1 of 1 Analysing: 100%|=========================================================| 151/151 [00:00<00:00, 512.43frame/s, rate=0] [FACEFUSION.CORE] Extracting frames with a resolution of 780x780 and 29.97002997002997 frames per second [FACEFUSION.FACE_SWAPPER] Processing Processing: 13%| | 19/151 [00:16<01:56, 1.14frame/s, execution_providers=['cpu'], execution_thread_count=4, execution [FACEFUSION.CORE] Processing stopped [FACEFUSION.CORE] Processing step 1 of 1 [FACEFUSION.CORE] Extracting frames with a resolution of 780x780 and 29.97002997002997 frames per second [FACEFUSION.FACE_SWAPPER] Processing Processing: 100%|=| 151/151 [00:11<00:00, 13.63frame/s, execution_providers=['cuda'], execution_thread_count=4, executi [FACEFUSION.CORE] Merging video with a resolution of 780x780 and 29.97002997002997 frames per second [FACEFUSION.CORE] Processing to video succeed in 14.24 seconds Exception in callback _ProactorBasePipeTransport._call_connection_lost(None) handle: <Handle _ProactorBasePipeTransport._call_connection_lost(None)> : ConnectionResetError: [WinError 10054] 既存の接続はリモート ホストに強制的に切断されました。