Object Detection Api

Abstract

Object detections is a common computer vision task used in most common applications. This project leverages a combination of Oat++, OpenCV, and YOLOv8 over c++ 17 to create a robust Object Detection API. The objective of this repository is make a scalable object detections api build in c++17, to do it, this project use Docker and cmake to ease the development process.

What is OAT++
What is OpenCV
What is Yolo and Yolo v8?
Setup
Test
References

What is oat++?

Oat++ is a C++ web framework designed to build high-performance and scalable web applications. It is open-source and primarily focuses on providing developers with a simple and efficient way to create RESTful APIs (Representational State Transfer) and web services.

Important Notes:

Compatibility with SQLite, PostgreSQL and MongoDB.
High demand supported (5 million concurrent).
Zero Dependency.

What is OpenCV?

Note that actually this project is tested using opencv 4.8.1.

It is a open source computer vision library designed to provide a comprehensive set of tools for image and video processing. Developed in C++ and with bindings available for various programming languages, including Python, javascript and java. OpenCV is widely used in both academic and industrial settings for a range of applications. Its features include image and video manipulation, object detection, facial recognition, feature extraction, camera calibration, and machine learning integration.

What is Yolo and Yolo v8?

YOLO (You Only Look Once) is a popular object detection model known for its speed and accuracy. It is a model designed based on SSD architecture. With the passage some group of research was improved this architecture to make faster and update the precision.

Yolov8 is an implementation by Ultralytics group.

SetUP

To run this project there are two way to compile the solve, the best way to compatibility is using docker, and the second is build the solution using cmake as package involve solution.

Download the model

conda create --name object_detection python=3.11 -y
conda activate object_detection
pip install ultralytics
yolo export model=yolov8n.pt format=onnx imgsz=640,640

it almost supports yolo v5 only make sure that image size is 640.

Cmake:

build the oat++ library, to see all cmake params, please go to this link

git clone https://github.com/oatpp/oatpp.git
cd oatpp
mkdir build && cd build
cmake --build . --config Release 
sudo make install

build the opencv library, will depend ind most cases of your hardware and software installed, but take consider that it project your the dnn component of opencv and it require to compile with ProtoBuf

If your are using macos and you are use brew as package manager, the las version 4.9.0 don't use protobuf, please install the last 4.8.1

brew install --build-from-source ./opencv.rb

Another SO

OPENCV_VERSION="4.8.1"
MODULE_NAME="opencv"

wget -O ${MODULE_NAME}.zip https://github.com/opencv/opencv/archive/${OPENCV_VERSION}.zip && \
wget -O ${MODULE_NAME}_contrib.zip https://github.com/opencv/opencv_contrib/archive/${OPENCV_VERSION}.zip

unzip -o ${MODULE_NAME}.zip -d ${MODULE_NAME}
unzip -o ${MODULE_NAME}_contrib.zip -d ${MODULE_NAME}

cd ${MODULE_NAME}-${OPENCV_VERSION}

cmake -D CMAKE_BUILD_TYPE=RELEASE \
-D CMAKE_INSTALL_PREFIX=/usr/local \
-D OPENCV_EXTRA_MODULES_PATH=../../opencv_contrib-4.8.1/modules \
-D CMAKE_C_COMPILER=/usr/bin/clang \
-D CMAKE_CXX_COMPILER=/usr/bin/clang++ \
-D CMAKE_INSTALL_PREFIX=/usr/local \
-D INSTALL_PYTHON_EXAMPLES=OFF \
-D INSTALL_C_EXAMPLES=OFF \
-D WITH_FFMPEG=ON \
-D OPENJPEG_LIBRARIES=/usr/lib/libopenjp2.so \
-D WITH_TBB=ON \
..

Build the api

mkdir build && cd build  
cmake -D OpenCV_DIR=/usr/lib/cmake/opencv4 ..
make

Run api

./api_machine_learning-exe

Docker:

The docker image take up approximately 1.72GB using alpine base image and default opencv package distributed by pkg.
```
docker build object_detection_api:2 .
```
If you want to build opencv it's possibles you need ubuntu to better package compatibility, note that it work with amd processors 2.66GB.
```
docker build --platform=linux/amd64 -t object_detection_api:2 -f ./dockerfile_custom
```
```
docker run -d --name ObjectDetectionApi -p8000:8000 object_detection_api:2
```

Test

Request:

curl --location --request POST 'localhost:8000/images' \
--form 'image=@"image_file.jpg"'

References

[1] Jocher, G., Chaurasia, A., & Qiu, J. (2023). Ultralytics YOLO (Version 8.0.0) [Computer software]. https://github.com/ultralytics/ultralytics

[2] Bradski, G. (2000). The OpenCV Library. Dr. Dobb's Journal of Software Tools.

[3] Stryzhevskyi, L., & Mokroß, B.-A. (n.d.). OATPP/OATPP: 🌱light and powerful C++ web framework for highly scalable and resource-efficient web application. it’s zero-dependency and easy-portable. GitHub. https://github.com/oatpp/oatpp

[4] Kundu, R. (2023, January 17). Yolo algorithm for object detection explained [+examples]. YOLO Algorithm for Object Detection Explained [+Examples]. https://www.v7labs.com/blog/yolo-object-detection

DanielSarmiento04/ObjectDetectionApi