SOFTWARE

I work on software projects focused on computer vision, automation, and intelligent systems that interact with the real world. Many of my projects involve using cameras and machine learning to detect, track, and interpret objects in real-time.

This page includes a selection of my programming projects, experiments, and technical work both in-class and at home.

Current Project

VISION DRIVER

An efficient and versatile computer vision system designed to detect, track, and interpret objects from video and image input, while running on most modern personal computers.

Vision Driver is a real-time computer vision system designed to detect, track, and interpret objects using live camera input*. The goal of the project was to create an efficient and versatile program that can run on everyday hardware** while still providing accurate and responsive object detection.

The system uses machine learning models and image processing techniques to identify objects such as vehicles and pedestrians in real time. It can analyze live video feeds*, track movement across frames, and determine what is happening in a scene as it unfolds. This allows the system to be used in applications such as traffic monitoring, automation, safety systems, and smart environments.

A major focus of this project was efficiency and accessibility. Many computer vision systems require powerful and expensive hardware, but Vision Driver was designed to run on standard personal computers**, making it more practical and widely usable.

This project is part of my broader interest in intelligent systems that interact with the real world through cameras and sensors. I am particularly interested in how computer vision can be used to improve transportation, automation, and human-computer interaction.

Current System Capabilities

• Detects and identifies vehicles and pedestrians in video and image input.

• Tracks object movement across multiple frames.

• Processes pre-recorded video and image files.

• Designed for efficient performance on personal computers.

• Supports multiple object detection within a single scene.

• Built using machine learning and image processing techniques.

Recommended System Specifications

To ensure smooth performance and accurate object detection, the following system specifications are recommended when running Vision Driver:

Minimum Requirements

• Processor: Dual-core CPU, 2.0 GHz or higher

• RAM: 8 GB

• Graphics: Integrated graphics (Intel Iris / AMD Radeon integrated or similar)

• Storage: 2 GB available space

• Operating System: Windows 10/11, macOS, or Linux

• Input: Pre-recorded video or image files

Recommended Specifications

• Processor: Quad-core CPU, 3.0 GHz or higher

• RAM: 16 GB

• Graphics: Dedicated GPU recommended (NVIDIA GTX 1650 or better)

• Storage: 5 GB available space (SSD recommended)

• Operating System: Windows 11, macOS, or Linux

• Input: High-resolution video supported for better detection accuracy

Notes

Performance may vary depending on video resolution, the number of objects in the scene, lighting conditions, and overall hardware configuration. Systems with dedicated graphics processing units (GPUs) will typically experience significantly faster processing and smoother performance compared to systems using integrated graphics.

Vision Driver Test Devices

• Tested on: MacBook Air (M3), 16 GB RAM

• Tested on: Windows PC, AMD Ryzen 7 processor, 32 GB RAM, NVIDIA RTX 3070Ti

Average performance during testing ranged from approximately 15–25 frames per second (FPS), depending on video resolution, lighting conditions, and the number of objects present in the scene.***

*Vision Driver currently supports pre-recorded video and image inputs only. Live video processing is planned for a future update.

**Smooth performance and maximum detection accuracy are not guaranteed on older hardware or systems below the recommended specifications. Performance may vary depending on video resolution, lighting conditions, and scene complexity.

***Testing was performed using a pre-release build of Vision Driver. This version is not yet publicly available and is used for internal testing purposes only. Public releases may vary slightly in performance and features.