Computer Vision Toolkit

Overview

Six real-time computer vision modules built with OpenCV, covering edge detection, object tracking, and video stabilization on a live webcam feed.

The Problem

Most computer vision tutorials teach each technique in isolation. I wanted a modular toolkit where every module sits on shared foundations and you can test it live on a webcam.

My Role

Sole author. Designed the modular architecture and wrote each vision module on its own, over a shared set of utilities.

Key Features

• —Edge detection with configurable Canny thresholds
• —Real-time object tracking with bounding box rendering
• —Video stabilization for handheld/shaky footage
• —Modular architecture: each module runs on its own
• —All processing runs at <100ms latency on webcam feeds

Challenges

Holding frame rates across all modules at once. I profiled each pipeline and rewrote the NumPy operations that were creating per-frame bottlenecks.

What I Learned

Next time I'd measure per-module latency from the first commit instead of profiling after the frame rate dropped.