opencv-mobile

✔️ This project provides the minimal build of opencv library for the Android, iOS and ARM Linux platforms.

✔️ Packages for Windows, Linux, MacOS, HarmonyOS and WebAssembly are available now.

✔️ We provide prebuild binary packages for opencv 2.4.13.7, 3.4.20 and 4.10.0.

✔️ We also provide prebuild package for Mac-Catalyst, watchOS, tvOS, visionOS and Apple xcframework.

✔️ All the binaries are compiled from source on github action, no virus, no backdoor, no secret code.

✔️ NEW FEATURE cv::putText supports full-width CJK characters

✔️ NEW FEATURE cv::imshow supports Linux framebuffer and Windows

opencv 4.10.0 package size	The official opencv	opencv-mobile
source zip	95.2 MB	8.25 MB
android	292 MB	17.7 MB
ios	207 MB	3.97 MB

Technical Exchange QQ Group 623504742（opencv-mobile夸夸群）

Download

https://github.com/nihui/opencv-mobile/releases/latest

	Source
	Android
	HarmonyOS
	iOS
	iOS-Simulator
	macOS
	Mac-Catalyst
	watchOS
	watchOS-Simulator
	tvOS
	tvOS-Simulator
	visionOS
	visionOS-Simulator
	Apple xcframework
	Ubuntu-20.04
	Ubuntu-22.04
	VS2015
	VS2017
	VS2019
	VS2022
	WebAssembly
	ARM-Linux

• Android package build with ndk r26c and android api 21
• iOS / MacOS / Mac-Catalyst / watchOS / tvOS / visionOS package build with Xcode 15.2
• ARM Linux package build with cross-compiler on Ubuntu-22.04
• WebAssembly package build with Emscripten 3.1.28

opencv-mobile package for development boards

milkv-duo riscv64-linux-musl ✅ HW JPG decoder ✅ MIPI CSI camera	licheerv-nano riscv64-linux-musl ✅ HW JPG decoder ✅ MIPI CSI camera	luckfox-pico arm-linux-uclibcgnueabihf ✅ HW JPG encoder ✅ MIPI CSI camera ✅ DPI LCD screen
yuzuki-lizard arm-linux-uclibcgnueabihf	tinyvision arm-linux-uclibcgnueabihf ✅ HW JPG decoder ✅ HW JPG encoder ✅ MIPI CSI camera ✅ SPI LCD screen	yuzuki-chameleon arm-openwrt-linux-gnueabi ✅ HW JPG decoder ✅ HW JPG encoder
purple-pi arm-linux-uclibcgnueabihf	myir-t113i arm-linux-gnueabi ✅ HW JPG decoder ✅ HW JPG encoder	2k0300-fengniao loongarch64-linux-gnu
lockzhiner-vision-module arm-linux-uclibcgnueabihf ✅ HW JPG encoder ✅ MIPI CSI camera ✅ DPI LCD screen

Usage Android

1. Extract archive to <project dir>/app/src/main/jni/
2. Modify <project dir>/app/src/main/jni/CMakeLists.txt to find and link opencv

set(OpenCV_DIR ${CMAKE_SOURCE_DIR}/opencv-mobile-4.10.0-android/sdk/native/jni)
find_package(OpenCV REQUIRED)

target_link_libraries(your_jni_target ${OpenCV_LIBS})

Usage iOS, macOS, watchOS, tvOS, visionOS

1. Extract archive, and drag opencv2.framework or opencv2.xcframework into your project

Usage ARM Linux, Windows, Linux, WebAssembly

1. Extract archive to <project dir>/
2. Modify <project dir>/CMakeLists.txt to find and link opencv
3. Pass -DOpenCV_STATIC=ON to cmake option for windows build

set(OpenCV_DIR ${CMAKE_SOURCE_DIR}/opencv-mobile-4.10.0-armlinux/arm-linux-gnueabihf/lib/cmake/opencv4)
find_package(OpenCV REQUIRED)

target_link_libraries(your_target ${OpenCV_LIBS})

How-to-build your custom package

We reduce the binary size of opencv-mobile in 3 ways

1. Reimplement some modules (such as highgui) and functions (such as putText)
2. Apply patches to disable rtti/exceptions and do not install non-essential files
3. Carefully select cmake options to retain only the modules and functions you want

Steps 1 and 2 are relatively cumbersome and difficult, and require intrusive changes to the opencv source code. If you want to know the details, please refer to the steps in .github/workflows/release.yml

The opencv-mobile source code package is the result of steps 1 and 2. Based on it, we can adjust the cmake option to compile our own package and further delete and add modules and other functions.

step 1. download opencv-mobile source

wget -q https://github.com/nihui/opencv-mobile/releases/latest/download/opencv-mobile-4.10.0.zip
unzip -q opencv-mobile-4.10.0.zip
cd opencv-mobile-4.10.0

step 2. apply your opencv option changes to options.txt

vim options.txt

step 3. build your opencv package with cmake

mkdir -p build
cd build
cmake -DCMAKE_INSTALL_PREFIX=install \
  -DCMAKE_BUILD_TYPE=Release \
  `cat ../options.txt` \
  -DBUILD_opencv_world=OFF ..
make -j4
make install

step 4. make a package

zip -r -9 opencv-mobile-4.10.0-mypackage.zip install

Some notes

• The minimal opencv build contains most basic opencv operators and common image processing functions, with some handy additions like keypoint feature extraction and matching, image inpainting and opticalflow estimation.

• Many computer vision algorithms that reside in dedicated modules are discarded, such as face detection etc. You could try deep-learning based algorithms with neural network inference library optimized for mobile.

• Image IO functions in highgui module, like cv::imread and cv::imwrite, are re-implemented using stb for smaller code size. GUI functions, like cv::imshow, are discarded.

• cuda and opencl are disabled because there is no cuda on mobile, no opencl on ios, and opencl on android is slow. opencv on gpu is not suitable for real productions. Write metal on ios and opengles/vulkan on android if you need good gpu acceleration.

• C++ RTTI and exceptions are disabled for minimal build on mobile platforms and webassembly build. Be careful when you write cv::Mat roi = image(roirect); :P

cv::putText supports full-width CJK characters

1. Open https://nihui.github.io/opencv-mobile/patches/fontface.html or opencv-mobile-X.Y.Z/fontface.html in your browser.
2. In the opened page, enter all the text to be drawn, select the TTF font file (optional), click the Convert to Font Header button to download the fontface header. This step is completely local operation, without connecting to a remote server, your data is private and safe.
3. Include the generated fontface header, initialize a fontface instance, and pass it as the argument to cv::putText. The source file must be encoded in UTF-8.

Since all characters have been converted to embedded bitmap, the drawing routine does not depend on freetype library or any font files at runtime.

#include <opencv2/core/core.hpp>
#include <opencv2/highgui/highgui.hpp>
#include <opencv2/imgproc/imgproc.hpp>
#include "myfontface.h"

int main()
{
    cv::Mat bgr = cv::imread("atari.jpg", 1);

    // use this font
    MyFontFace myfont;

    // draw full-width text with myfont
    const char* zhtext = "称呼机器人为破铜烂铁，\n违反了禁止歧视机器人法！";
    cv::putText(bgr, zhtext, cv::Point(30, 250), cv::Scalar(127, 0, 127), myfont, 20);

    // get bounding rect
    cv::Rect rr = cv::getTextSize(bgr.size(), zhtext, cv::Point(30, 250), myfont, 20);

    cv::imwrite("out.jpg", bgr);

    return 0;
}

cv::imshow supports Linux framebuffer and Windows

In Linux, cv::imshow can display images on the screen (/dev/fb0) via the Linux Framebuffer API. cv::imshow can work without desktop environment (gnome, KDE Plasma, xfce, etc.) or window manager (X or wayland), making it suitable for embedded scenarios. The first argument to cv::imshow must be fb.

In Windows, cv::imshow will use the Windows API to create a simple window for displaying.

display image #include <opencv2/core/core.hpp> #include <opencv2/highgui/highgui.hpp> int main() { cv::Mat bgr = cv::imread("im.jpg", 1); cv::imshow("fb", bgr); return 0; }

realtime camera preview #include <opencv2/core/core.hpp> #include <opencv2/highgui/highgui.hpp> int main() { cv::VideoCapture cap; cap.set(cv::CAP_PROP_FRAME_WIDTH, 320); cap.set(cv::CAP_PROP_FRAME_HEIGHT, 240); cap.open(0); cv::Mat bgr; while (1) { cap >> bgr; cv::imshow("fb", bgr); } return 0; }

opencv modules included

module	comment
opencv_core	Mat, matrix operations, etc
opencv_imgproc	resize, cvtColor, warpAffine, etc
opencv_highgui	imread, imwrite
opencv_features2d	keypoint feature and matcher, etc (not included in opencv 2.x package)
opencv_photo	inpaint, etc
opencv_video	opticalflow, etc

opencv modules discarded

module	comment
opencv_androidcamera	use android Camera api instead
opencv_calib3d	camera calibration, rare uses on mobile
opencv_contrib	experimental functions, build part of the source externally if you need
opencv_dnn	very slow on mobile, try ncnn for neural network inference on mobile
opencv_dynamicuda	no cuda on mobile
opencv_flann	feature matching, rare uses on mobile, build the source externally if you need
opencv_gapi	graph based image processing, little gain on mobile
opencv_gpu	no cuda/opencl on mobile
opencv_imgcodecs	link with opencv_highgui instead
opencv_java	wrap your c++ code with jni
opencv_js	write native code on mobile
opencv_legacy	various good-old cv routines, build part of the source externally if you need
opencv_ml	train your ML algorithm on powerful pc or server
opencv_nonfree	the SIFT and SURF, use ORB which is faster and better
opencv_objdetect	HOG, cascade detector, use deep learning detector which is faster and better
opencv_ocl	no opencl on mobile
opencv_python	no python on mobile
opencv_shape	shape matching, rare uses on mobile, build the source externally if you need
opencv_stitching	image stitching, rare uses on mobile, build the source externally if you need
opencv_superres	do video super-resolution on powerful pc or server
opencv_ts	test modules, useless in production anyway
opencv_videoio	use android MediaCodec or ios AVFoundation api instead
opencv_videostab	do video stablization on powerful pc or server
opencv_viz	vtk is not available on mobile, write your own data visualization routines