Computer Vision

Period: 2019 – 2020
Scope: Classical Computer Vision & Early Deep Learning
Tools: OpenCV · NumPy · Python

Overview

Worked on classical computer vision algorithms for detection, tracking, and motion analysis, while beginning to integrate early deep learning models into production workflows.

This role focused on:

• feature-based detection and tracking
• motion estimation and scene understanding
• transitioning from traditional CV → deep learning pipelines

Key Contributions

• Built real-time object detection and tracking systems
• Developed motion analysis and change detection pipelines
• Designed feature-based matching and geometric estimation algorithms
• Integrated deep learning models (YOLO) into practical applications
• Delivered end-to-end desktop-based CV applications

Works

• Change Detection
• Object Detection & Tracking
• Feature Matching & Image Registration
• Video Stabilization & Motion Estimation
• Horse Video Auto Clipper

Tech Focus

OpenCV · NumPy · Feature Engineering · Tracking · Geometry (RANSAC) · YOLO · ByteTrack

Impact

• Delivered robust real-world CV systems under noisy conditions
• Bridged classical CV and deep learning approaches
• Built reusable pipelines for tracking, detection, and motion analysis
• Established strong foundation for later deep learning-based systems