Apra-Labs/ApraPipes

Apra-Labs/ApraPipes has 37 GitHub stars and 38 forks. GitHot measured 0 stars in 24 hours and +2 in 7 days. Last checked 2026-11-10 03:14 UTC. Last push 2026-10-29.

How these counts are measured

A pipeline framework for developing video and image processing application. Supports multiple GPUs and Machine Learning tooklits

37 stars 38 forks

Measured star growth

in 24h: 0
in 7 days: +2
in 15 days: Update delayed
in 1 month: Update delayed

Net star changes over overlapping periods, not daily totals. Missing measurements are left blank. Stars measure attention, not software quality.

README.md

Original

Header Image

ApraPipes

A pipeline framework for developing video and image processing applications. Supports multiple GPUs and Machine Learning toolkits. Learn more about ApraPipes here.

Declarative Pipelines & Node.js Support

Build video processing pipelines using JSON configuration or JavaScript:

{
  "modules": {
    "source": { "type": "FileReaderModule", "props": { "path": "video.mp4" } },
    "decoder": { "type": "Mp4ReaderSource" },
    "sink": { "type": "FileSinkModule" }
  },
  "connections": [
    { "from": "source", "to": "decoder" },
    { "from": "decoder", "to": "sink" }
  ]
}

Features:

50+ registered modules (sources, transforms, encoders, AI inference)
GPU acceleration with automatic CPU↔GPU memory bridging
Node.js addon for JavaScript/TypeScript applications
CLI tool for running and validating pipelines

See examples/ and Pipeline Author Guide to get started.

Build Status

ApraPipes is automatically built and tested on Windows, Linux (x64 and ARM64), and macOS.

Platform	OS	Version	Test Results
x86_64	Windows	2022	NOCUDA: CUDA:
x86_64	Linux	22.04	NOCUDA: CUDA: Docker:
x86_64	macOS	15.0+	NOCUDA:
ARM64	Linux (Jetson)	20.04	CUDA:

Getting Started with ApraPipes

Please select your Operating System

Note : Make sure to clone using recursive flag

git clone --recursive https://github.com/Apra-Labs/ApraPipes.git

Windows (Version ≥ 10)

Requirements

Prerequisites

Visual Studio

Install Visual Studio 2019 Community
- Install Desktop development C++
- .NET Desktop development
- Universal Windows Development Platform

Cuda

Create an account on developer.nvidia.com if you're not already a member. Note : Otherwise the next step will show HTTP 404/403 error.
Windows 10/11 : Cuda Toolkit 10.2 or CUDA Toolkit 11.8.

Cudnn

Download Cudnn and extract files where cuda is installed. Note: Please be aware that this process requires some effort. Here are the necessary steps:
- Download the correct zip file matching your cuda version. Do not download the exe/installer/deb package.
- Windows:
  - Download this file.

Clone with submodules and LFS.

git clone --recursive https://github.com/Apra-Labs/ApraPipes.git

Build

Open PowerShell as an administrator and execute the following commands

Build Without Cuda

If your windows system does not have an NVIDIA GPU use this script

build_windows_no_cuda.bat

Build With Cuda

build_windows_cuda.bat

To Build With Documentation

build_windows_cuda.bat --build-doc

Test

Run Tests

list all tests

_build/BUILD_TYPE/aprapipesut.exe --list_content

run all tests
```
_build/BUILD_TYPE/aprapipesut.exe
```

run all tests disabling memory leak dumps and better progress logging

_build/BUILD_TYPE/aprapipesut.exe -p -l all --detect_memory_leaks=0

run one test

_build/BUILD_TYPE/aprapipesut.exe --run_test=filenamestrategy_tests/boostdirectorystrategy

run one test with arguments
```
_build/BUILD_TYPE/aprapipesut.exe --run_test=unit_tests/params_test -- -ip 10.102.10.121 -data ArgusCamera
```
- Look at the unit_tests/params_test to check for sample usage of parameters in test code.

Linux (Ubuntu ≥ 20.04)

Requirements

Prerequisites

Cuda

Create an account on developer.nvidia.com if you're not already a member. Note : Otherwise the next step will show HTTP 404/403 error.
Ubuntu 20.04/22.04:
- 20.04 - CUDA Toolkit 11.8
- 22.04 - CUDA Toolkit 12.x

Cudnn

Download cuDNN matching your CUDA version
Linux:
- For CUDA 11.8: cuDNN 8.x
- For CUDA 12.x: cuDNN 9.x

Clone with submodules and LFS.

git clone --recursive https://github.com/Apra-Labs/ApraPipes.git

Build

Run this command to make the script file executable.

chmod +x build_linux_*.sh

Build Without Cuda

If your windows system does not have an NVIDIA GPU use this script

./build_linux_no_cuda.sh

Build With Cuda

./build_linux_cuda.sh

To Build With Documentation

./build_linux_cuda.sh --build-doc

Build can take ~2 hours depending on the machine configuration.

Test

Run Tests

list all tests
```
./_build/aprapipesut --list_content
```
run all tests
```
./_build/aprapipesut
```
run all tests disabling memory leak dumps and better progress logging
```
./_build/aprapipesut -p -l all --detect_memory_leaks=0
```

run one test

./_build/aprapipesut --run_test=filenamestrategy_tests/boostdirectorystrategy

run one test with arguments
```
./_buildaprapipesut --run_test=unit_tests/params_test -- -ip 10.102.10.121 -data ArgusCamera
```
- Look at the unit_tests/params_test to check for sample usage of parameters in test code.

macOS (Version ≥ 15.0)

Requirements

Prerequisites

Homebrew

Install Homebrew package manager if not already installed:

/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

Build Tools

Install nasm (required for FFmpeg):
```
brew install nasm
```

Clone with submodules and LFS.

git clone --recursive https://github.com/Apra-Labs/ApraPipes.git

Build

Run this command to make the script file executable.

chmod +x build_macos.sh

Build Without Cuda (Only NoCUDA Build)

macOS builds do not support CUDA. Use the following script:

./build_macos.sh

Build can take ~2 hours depending on the machine configuration.

Test

Run Tests

list all tests
```
./build/aprapipesut --list_content
```
run all tests
```
./build/aprapipesut
```
run all tests disabling memory leak dumps and better progress logging
```
./build/aprapipesut -p -l all --detect_memory_leaks=0
```

run one test

./build/aprapipesut --run_test=filenamestrategy_tests/boostdirectorystrategy

run one test with arguments
```
./build/aprapipesut --run_test=unit_tests/params_test -- -ip 10.102.10.121 -data ArgusCamera
```
- Look at the unit_tests/params_test to check for sample usage of parameters in test code.

Jetson Boards - Nano, TX2, NX, AGX, Orin (JetPack ≥ 5.0)

Requirements

Prerequisites

Setup the board with JetPack 5.0+:
- JetPack 5.x: Ubuntu 20.04, CUDA 11.4
- JetPack 6.x: Ubuntu 22.04, CUDA 12.x

Clone with submodules and LFS.

git clone --recursive https://github.com/Apra-Labs/ApraPipes.git

Build

Build for Jetson (Only Cuda Build)

Run this command to make the script file executable.

chmod +x build_jetson.sh

ApraPipes builds CUDA version on Jetson Boards.

./build_jetson.sh

To Build With Documentation

./build_jetson.sh --build-doc

Build can take ~12 hours on Jetson Nano. Note: Jetson build can also be done using Ubuntu 20.04+ x86_64 Laptop via cross compilation.

Cross Compilation using qemu

Cross compilation using qemu

Conceptual steps adapted from here:

On any Intel Ubuntu 20.04+ computer (physical or virtual including wsl) mount a Jetson SD Card Image as described above
Copy relevant files from mounted image to created a rootfs
Install qemu on ubuntu host
chroot into emulated aarm64 environment using script provided in the github link above
install extra tools and build aprapipes and aprapipesut
the built aprapipesut can be copied to a Jetson board and run.

This approach can use all 12-16 cores of a laptop and hence builds faster.

Test

Run Tests

list all tests ./_build/aprapipesut --list_content
run all tests ./_build/aprapipesut
run one test ./_build/aprapipesut --run_test=filenamestrategy_tests/boostdirectorystrategy
run one test with arguments ./_build/aprapipesut --run_test=unit_tests/params_test -- -ip 10.102.10.121 -data ArgusCamera
Look at the unit_tests/params_test to check for sample usage of parameters in test code

Docker

Requirements

Prerequisites

Ensure virtualization is enabled in both the BIOS settings of your computer and the Windows virtualization feature -Refer this article to enable them
Install WSL 2 on your system:
```
wsl --install
```
Set WSL 2 as the default version using the command line:
```
wsl --set-default-version 2
```
Install Ubuntu 22.04 from Microsoft Store or run wsl --install -d Ubuntu-22.04
Install Docker Desktop on Windows from here
Enable Docker integration with WSL 2 (Docker Desktop settings → Resources → WSL integration → Enable Ubuntu-22.04 → Apply & Restart)
Install nvidia-container-toolkit in WSL Ubuntu for GPU access - Follow this guide
Note:"Follow the exact instructions outlined in the document to ensure the correct and successful installation of the NVIDIA Container Toolkit"

Build

Build for Docker

Use this docker image with all dependencies pre-installed:

docker pull ghcr.io/kumaakh/aprapipes-build-x86-ubutu18.04-cuda:last-good

Note: This image is based on Ubuntu 18.04 with pre-cached vcpkg dependencies for fast builds (~10 min). The WSL host can be Ubuntu 22.04.

Mount an external volume as a build area, and then use the Windows command line to create a Docker container using the above image with the following command:
```
docker run -dit --gpus all -v "</path/to/external_volume>":"/mnt/b/" --name <give-container-name> a799cc26f4b7
```
..your command should look like this [where D:\ws\docker-pipes->local_folder_path , pipes->container_name ]
```
docker run -dit --gpus all -v "D:\ws\docker-pipes":"/mnt/b/" --name pipes a799cc26f4b7
```
After creating the container, execute the following command to access its command line interface
```
docker exec -it <container-name> /bin/bash
```
Note:"When inside the container, build all contents within the mounted external folder"
clone the repository with submodules and LFS as described above
build using build_linux_*.sh scripts as described above

This build will be fairly fast (~10 mins) as entire vcpkg cache comes down with the docker image

Update Submodules

git submodule update --init --recursive

Update Documentation

To update documentation, refer to Documentation Guidelines in the Contribution-Guidelines.

To regenerate documentation

Run,

./build_documentation.sh

Measured star growth

README.md

ApraPipes

Declarative Pipelines & Node.js Support

Build Status

Getting Started with ApraPipes

Windows (Version ≥ 10)

Prerequisites

Visual Studio

Cuda

Cudnn

Build Without Cuda

Build With Cuda

To Build With Documentation

Run Tests

Linux (Ubuntu ≥ 20.04)

Prerequisites

Cuda

Cudnn

Build Without Cuda

Build With Cuda

To Build With Documentation

Run Tests

macOS (Version ≥ 15.0)

Prerequisites

Homebrew

Build Tools

Build Without Cuda (Only NoCUDA Build)

Run Tests

Jetson Boards - Nano, TX2, NX, AGX, Orin (JetPack ≥ 5.0)

Prerequisites

Build for Jetson (Only Cuda Build)

Cross compilation using qemu

Run Tests

Docker

Prerequisites

Build for Docker

Update Submodules

Update Documentation

To regenerate documentation

Explore related projects

ytdl-org/youtube-dl

excalidraw/excalidraw

godotengine/godot

mrdoob/three.js