Skip to main content

rmbg-video

使用 rembg AI 模型逐帧去除视频背景,输出带透明通道的 VP8 WebM 视频,保留原音频。

系统要求

  • Python >= 3.9
  • ffmpeg 和 ffprobe 已安装并在 PATH 中

安装

pip install rmbg-video

使用方式

命令行

# 基本用法
rmbg-video input.mp4
rmbg-video input.mp4 output.webm

# 选择模型
rmbg-video input.mp4 --model u2net

# 调节质量 (0-63,越小越好)
rmbg-video input.mp4 --crf 5

# 输出不含透明通道的普通 WebM
rmbg-video input.mp4 --no-alpha

# 跳过音频处理
rmbg-video input.mp4 --no-audio

# 强制使用 CPU
rmbg-video input.mp4 --no-gpu

# 保留临时文件用于调试
rmbg-video input.mp4 --keep-temp

# 测试模式:只处理前5帧(快速预览效果)
rmbg-video input.mp4 --test

Web 界面

启动 Web 服务后,团队成员可通过浏览器上传视频、配置参数并预览处理结果。处理任务自动排队,单 GPU 串行执行。

# 启动 Web 服务
rmbg-video-web

# 或通过 Python 模块启动
python -m rmbg_video.web

启动后访问 http://localhost:7860,界面功能:

  • 上传视频 — 支持 .mp4 / .webm / .mov / .avi / .mkv 格式
  • 参数配置 — 所有 CLI 参数均可通过界面调整,测试模式(仅处理前 5 帧)快捷开关位于最前
  • 排队处理 — 多人同时提交时自动排队,显示队列状态,单任务串行处理
  • 透明预览 — 结果视频以灰色棋盘格为背景预览,直观查看透明通道效果
  • 下载结果 — 处理完成后可直接下载 WebM 视频

选项

选项 默认值 说明
--model birefnet-general rembg 模型名
--no-alpha-matting 禁用 alpha matting(更快但边缘较硬)
--fg-threshold 240 前景阈值
--bg-threshold 10 背景阈值
--erode-size 10 腐蚀尺寸
--post-process-mask 对遮罩做后处理平滑
--crf 10 VP8 质量 0-63(越小越好)
--speed best 编码速度预设 (good/best/realtime)
--no-alpha 输出不含透明通道的普通 WebM
--no-audio 跳过音频处理
--keep-temp 保留临时文件用于调试
--ffmpeg-path 自动检测 ffmpeg 可执行文件路径
--no-gpu 强制使用 CPU
--test 测试模式:只处理前5帧

工作原理

  1. 使用 ffmpeg 管道流式解码视频为原始 RGBA 帧
  2. 通过 rembg AI 模型逐帧去除背景,应用 alpha matting 平滑边缘
  3. 通过 ffmpeg 管道编码为 VP8 + yuva420p 像素格式(支持透明通道)
  4. 提取原音频为 Opus 编码,与视频流合并输出最终 WebM

GPU配置相关

  • 用的是cuda12.8版本

  • 要卸载 onnxruntime

    • uv pip uninstall onnxruntime onnxruntime-gpu onnxruntime-azure -y
    • uv pip install onnxruntime-gpu
  • 安装cuda

  • 安装cudnn

  • 根据错误提示把C:\Program Files\NVIDIA\CUDNN\v9.23\bin\12.9\x64,E:\AI\rmbg-video\.venv\Lib\site-packages\tensorrt_libs 等dll所在路径加入到系统path中

  • uv pip install tensorrt-cu12==10.9.0.34

许可证

MIT

Release files for rmbg-video 1.0.6

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for rmbg-video 1.0.6
File Size Uploaded
rmbg_video-1.0.6.tar.gz 22.3 kB Details

Built distribution (wheel)

Table of built distributions (wheels) for rmbg-video 1.0.6
File Interpreter ABI Platform
rmbg_video-1.0.6-py3-none-any.whl Python 3 none any Details

Total release size:35.0 kB

Release files / rmbg_video-1.0.6.tar.gz

Download URL rmbg_video-1.0.6.tar.gz
Size 22.3 kB
Tags Source
SHA-256 checksum
How to use checksums
760e379f35bfc9d18412d323c4f3001dfe06a79f3dde7bb347c23b8ec026179b
BLAKE2b-256 checksum
How to use checksums
9a16c469534b04c71ea8bcff6ac698d7c294f33d2660c05831b6964eb619753d
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/6.1.0 CPython/3.13.12

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 3, 2026.

Transparency log

Release files / rmbg_video-1.0.6-py3-none-any.whl

Download URL rmbg_video-1.0.6-py3-none-any.whl
Size 12.7 kB
Tags Python 3
SHA-256 checksum
How to use checksums
1cd417e00e63b189e6d69e3d09d524f138f84564e6925be5b92a4dfb7fc1c5a0
BLAKE2b-256 checksum
How to use checksums
4aabec89e6f015ed2bea218d2c0b6816bb1f8b447d32d3f02a97849a147b155d
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/6.1.0 CPython/3.13.12

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 3, 2026.

Transparency log

Release history Release notifications | RSS feed

This release

1.0.6 This release

2 release files

1.0.5

2 release files

1.0.4

2 release files

1.0.3

2 release files

1.0.2

2 release files

1.0.1

2 release files

1.0.0

2 release files

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page