projects

2026

1. Research

   Robust Text-to-Image Person Re-Identification for Real-World Search

   Hyukjin Kim

   2026

   Ongoing (Aug. 2026 – present)

   Abs

   • Text-based person retrieval for missing-person and suspect searches where reference images are unavailable and only natural-language descriptions may be provided
   • Adding image-grounded attributes generated by a vision-language model produced only modest gains, which led to asking which textual details the model actually uses
   • Controlled attribute-ablation experiments on IRRA across CUHK-PEDES, RSTPReID, and ICFG-PEDES: removing upper-body clothing descriptions reduced Rank-1 by 31.6, 25.6, and 43.3 percentage points; the relative importance of garment color and type differed across datasets

2. Research

   Multimodal Human Proficiency Estimation (SkillLens)

   Hyukjin Kim

   2026

   Apr. 2026 – Jun. 2026 · CVPR 2026 Workshop on MOMA (Oral)

   Abs Website

   • Studied how an action is performed, not just which action: whether local motion patterns capture proficiency differences that global video representations miss
   • Contributed core ideas and experimental design for action-guided Prototype Dynamics Margin (PDM) and CORAL-based ordinal calibration
   • Combined egocentric and exocentric video with gaze, using atomic-action descriptions as temporal guidance; 54.2% accuracy on Ego-Exo4D, outperforming the strongest evaluated prior baseline (50.1%) by 4.1 percentage points

3. Research

   Federated Domain Generalization for Person Re-Identification

   Hyukjin Kim

   2026

   Mar. 2026 – present · Motivated by expanding a deployed missing-person search system across municipal CCTV sites

   Abs

   • Investigated cross-domain generalization conflicts in federated person re-identification
   • Developed identity-ranking-based diagnostics in feature space
   • Leave-one-client-out experiments to assess individual client contributions to unseen-domain retrieval

4. Research

   Pose-Robust Representation Learning for Person Re-Identification (SwapMixer)

   Hyukjin Kim

   2026

   Dec. 2025 – Mar. 2026

   Abs

   • Designed and implemented SwapMixer, a generation-free feature-level augmentation method
   • Combines identity representations with pose embeddings from different identities
   • Removes the need for a separate pose-conditional image generator

2024

1. Capstone

   Improving Open-Vocabulary Object Detection via Knowledge Distillation

   Hyukjin Kim and Junghyun Park

   2024

   3rd place, Most Popular Award @ EE-Festival

   Abs PDF

   • Baseline: BARON (Aligning Bag of Regions for open-vocabulary object detection)
   • Identified a mismatch between teacher and student similarity structures for bags of regions sampled around the same region proposal
   • Applied similarity-preserving knowledge distillation to align teacher–student relational representations
   • Improved novel-class box AP50 on COCO by +1.2 points (23.1 → 24.3)

2. Project

   Self-Supervised Image Classification with SwAV

   Hyukjin Kim

   2024

   Abs PDF

   • Online clustering with the Sinkhorn–Knopp algorithm
   • Multi-view consistency on a ResNet-18 backbone
   • Transferred a pretext-trained network to CIFAR-100 classification

3. Coursework

   Deep Learning Lab — 12 Implementation Projects

   Hyukjin Kim

   2024

   Abs

   • VGG-16 and ResNet — CIFAR-10 classification
   • Spatial Transformer Network — distorted MNIST
   • FSRCNN — super-resolution
   • FCN — semantic segmentation
   • RetinaNet with Focal Loss — object detection
   • Grad-CAM — AlexNet and VGGNet visualization
   • Neural style transfer
   • DCGAN — image generation
   • CycleGAN — unpaired image translation
   • Seq2Seq with attention — machine translation
   • Network quantization — QIL with a Straight-Through Estimator
   • Self-supervised learning (SwAV) — CIFAR-100

2023

1. Coursework

   Foundations of AI — ML Algorithms from Scratch

   Hyukjin Kim

   2023

   Abs

   • Lasso and Ridge regression
   • Decision Tree and AdaBoost classifiers
   • Modular neural networks (forward / backward passes)
   • Implemented without ML libraries