Overview

This is a first time attempt at object detection in C++. The main file is where the object detection is written. Within main it loads a YOLOv5 nano model using cv::dnn::readNet(...). Then runs inference on either a single image or on video feed, the user is given the option to pick. Once the inference is run, it maps the bounding boxes on the feed/input and then displays it.

I wrote this software to improve my general C++ skills and learn more about object detection inside this language. I learned about NMS, object detection workflow, some new OpenCV functions, and how to work with video feed.

Object Detection in C++

Development Environment

This project used Visual Studio and OpenCV to complete.

Useful Websites

• Model Repository
• Empty Image CV
• Medium Article

Future Work

These items need to be fixed in future updates.

• Video feed needs to update overtime.
• Implement this into command line arguments.
• Add more comments to code explaining each step.