ViGent/Docs/DEPLOY_MANUAL.md

ViGent 手动部署指南

服务器信息

配置	规格
服务器	Dell PowerEdge R730
CPU	2× Intel Xeon E5-2680 v4 (56 线程)
内存	192GB DDR4
GPU 0	NVIDIA RTX 3090 24GB
GPU 1	NVIDIA RTX 3090 24GB (用于 MuseTalk)
部署路径	/home/rongye/ProgramFiles/ViGent

---

步骤 1: 环境检查

# 检查 GPU
nvidia-smi

# 检查 Python 版本 (需要 3.10+)
python3 --version

# 检查 Node.js 版本 (需要 18+)
node --version

# 检查 FFmpeg
ffmpeg -version

如果缺少 FFmpeg:

sudo apt update
sudo apt install ffmpeg

---

步骤 2: 创建目录结构

mkdir -p /home/rongye/ProgramFiles/ViGent
cd /home/rongye/ProgramFiles/ViGent

将项目文件复制到该目录。

---

步骤 3: 安装后端依赖

cd /home/rongye/ProgramFiles/ViGent/backend

# 创建虚拟环境
python3 -m venv venv
source venv/bin/activate

# 安装 PyTorch (CUDA 12.1)
pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121

# 安装其他依赖
pip install -r requirements.txt

# 安装 Playwright 浏览器 (社交发布用)
playwright install chromium

---

步骤 4: 安装 MMPose (唇形检测)

source /home/rongye/ProgramFiles/ViGent/backend/venv/bin/activate

pip install -U openmim
mim install mmengine
mim install "mmcv>=2.0.1"
mim install "mmdet>=3.1.0"
mim install "mmpose>=1.1.0"

---

步骤 5: 安装 MuseTalk

cd /home/rongye/ProgramFiles/ViGent/models

# 克隆仓库
git clone https://github.com/TMElyralab/MuseTalk.git
cd MuseTalk

# 激活虚拟环境
source /home/rongye/ProgramFiles/ViGent/backend/venv/bin/activate

# 安装依赖
pip install -r requirements.txt

---

步骤 6: 下载 MuseTalk 模型权重

从 HuggingFace 下载模型:

• 地址: https://huggingface.co/TMElyralab/MuseTalk

cd /home/rongye/ProgramFiles/ViGent/models/MuseTalk

# 使用 huggingface-cli 下载 (需要安装 huggingface_hub)
pip install huggingface_hub
huggingface-cli download TMElyralab/MuseTalk --local-dir ./models

或手动下载后放到:

/home/rongye/ProgramFiles/ViGent/models/MuseTalk/models/

---

步骤 7: 配置环境变量

cd /home/rongye/ProgramFiles/ViGent/backend

# 复制配置模板
cp .env.example .env

# 编辑配置
nano .env

修改以下配置:

# GPU 配置
MUSETALK_GPU_ID=1
MUSETALK_LOCAL=true

# 其他配置按需修改
DEBUG=false

---

步骤 8: 安装前端依赖

cd /home/rongye/ProgramFiles/ViGent/frontend

# 安装依赖
npm install

---

步骤 9: 测试运行

启动后端

cd /home/rongye/ProgramFiles/ViGent/backend
source venv/bin/activate
uvicorn app.main:app --host 0.0.0.0 --port 8000

启动前端 (新开终端)

cd /home/rongye/ProgramFiles/ViGent/frontend
npm run dev -- --host 0.0.0.0

---

步骤 10: 验证

1. 访问 http://服务器IP:3000 查看前端
2. 访问 http://服务器IP:8000/docs 查看 API 文档
3. 上传测试视频，生成口播视频

---

使用 systemd 管理服务 (可选)

后端服务

创建 /etc/systemd/system/vigent-backend.service:

[Unit]
Description=ViGent Backend API
After=network.target

[Service]
Type=simple
User=rongye
WorkingDirectory=/home/rongye/ProgramFiles/ViGent/backend
Environment="PATH=/home/rongye/ProgramFiles/ViGent/backend/venv/bin"
ExecStart=/home/rongye/ProgramFiles/ViGent/backend/venv/bin/uvicorn app.main:app --host 0.0.0.0 --port 8000
Restart=always

[Install]
WantedBy=multi-user.target

前端服务

创建 /etc/systemd/system/vigent-frontend.service:

[Unit]
Description=ViGent Frontend
After=network.target

[Service]
Type=simple
User=rongye
WorkingDirectory=/home/rongye/ProgramFiles/ViGent/frontend
ExecStart=/usr/bin/npm run start
Restart=always

[Install]
WantedBy=multi-user.target

启用服务

sudo systemctl daemon-reload
sudo systemctl enable vigent-backend vigent-frontend
sudo systemctl start vigent-backend vigent-frontend

---

故障排除

GPU 不可用

# 检查 CUDA
nvidia-smi
python3 -c "import torch; print(torch.cuda.is_available())"

端口被占用

# 查看端口占用
sudo lsof -i :8000
sudo lsof -i :3000

查看日志

# 后端日志
journalctl -u vigent-backend -f

# 前端日志
journalctl -u vigent-frontend -f