Gavin Joo

I am an Undergraduate Research Student in the Department of Arts & Technology, Computer Science & Engineering at Chung-Ang University (CAU), where I am fortunate to be advised by Prof. Sohn. My research interests lie in Computer Vision, Action Recognition, and Multimodal Learning, particularly focused on video data refinement and Large Language Models (LLMs).

I explore the intersection of A&T and Software Engineering, focusing on Visual Intelligence that prioritizes user-centric experiences. My research interests lie in building large-scale video data pipelines and advancing Action Recognition technologies. I expect to complete my B.S. in August 2027.

[ Email1 Email2 / LinkedIn / ]
Gavin Joo

Education

Research & Projects

Action Recognition Research
Development of VR Rhythm Game Interface Using Deep Learning-Based Gesture Recognition
Gavin Joo, Advised by Prof. Sohn
Undergraduate Research Program, CAU, 2025-2026

Integrated Soft-DTW loss into VideoPrism and TimeSformer to enhance temporal sensitivity in gesture recognition.
Managed large-scale experiments on Kinetics-400 and SSv2 (200K+ videos) using Google Cloud TPU v4-8 and NVIDIA A100.

[GitHub]
Techmals Project
AI-driven Identification System for Missing Animals
Gavin Joo
Ongoing Project, 2024 - present

Developed an individual identification system for stray animals using Deep Metric Learning and EfficientNet.
Integrated Triplet Loss to calculate similarity between missing reports and shelter databases via Gradio UI.

[GitHub]
IoT Food Spoilage Detection
Preliminary Study on Food Spoilage Anomaly Detection via Threshold Analysis of IoT CO₂ Sensor Time-Series Data
Gavin Joo, Sinyoung Jang
Preliminary Research Paper, 2022

Developed a low-cost IoT monitoring system using Arduino and SGP30 sensors to collect real-time eCO₂ time-series data.
Proposed a rule-based anomaly detection threshold (1,000ppm) by analyzing CO₂ emission patterns of decomposing food samples.

[Paper]