cv

Bachelor of Engineering in Computer Engineering

Undergraduate Research Assistant

• Developed and evaluated a hybrid augmentation pipeline combining FLUX.1-Fill, fractal-tree masks, and thermal-domain prompts to generate realistic vegetation-like occlusions.
• Trained YOLO11m on the augmented dataset, improving recall by 4.9 percentage points on the original test set and 12.3 points on the FLUX-occluded test set.

Data Augmentation for Improving Thermal Object Detection under Partial Occlusion

• Minjung Gong* · Suyoung Cho* · Jinsun Park
• * Equal contribution. Title translated; paper written in Korean.

Broom: On-Device AI Photo Management

• Led computer vision and machine learning development, covering dataset construction, model training, threshold calibration, and mobile inference.
• Consolidated five classifiers into one MobileNetV2 model and applied INT8 TensorFlow Lite quantization, reducing model size by approximately 75%.
• Built custom-album and duplicate-photo grouping with image embeddings and cosine similarity; registered the software with the Korea Copyright Commission.

Arctic Sea-Ice Forecasting

• Co-developed and submitted a concept for sea-ice forecasting and Arctic route planning using attention-based fusion of satellite imagery with oceanographic and meteorological variables.

Deepfake Detection AI Challenge 2025

• Led the design and curation of a synthetic-image dataset spanning demographic groups, ages, activities, occupations, lighting conditions, viewpoints, and aspect ratios.
• Built structured prompt libraries for multiple image-generation platforms and organized candidate real and synthetic data sources to broaden domain coverage.

Volunteer Korean Language Tutor

• Tutored international students in Korean language and culture and created customized learning materials for complex grammatical structures.