Jeho Lee | Yonsei University

About Me

I am a Ph.D. student at Yonsei University, working under the supervision of Prof. Hojung Cha. My research focuses on building real-time, on-device vision AI systems for mobile and edge platforms, enabling next-generation spatial computing applications such as AR/XR and autonomous systems. My core research interests include:

On-Device AI (with a primary focus on vision AI)
Mobile and Embedded Systems

To address the challenges of accelerating deep vision models on resource-constrained platforms, I take the following approaches:

AI-system co-optimization: jointly optimizing AI models and system software to meet real-time latency targets on given hardware
Heterogeneous computing on mobile SoCs: hardware-aware AI inference acceleration on heterogeneous processors (CPU/GPU/NPU), minimizing off-chip memory access and maximizing parallelism

News

[Aug. 2025] Two papers are accepted to ACM MobiCom 2025.
[Jul. 2025] Two papers are accepted to ACM EMSOFT 2025.
[June. 2025] Our paper won the Best Paper Award 🏆 at ACM MobiSys 2025!
[May. 2025] I am selected as one of Rising Stars at ACM MobiSys 2025 and received SIGMOBILE Student Travel Grant for attending MobiSys 2025.
[Apr. 2025] Our paper is accepted to ACM MobiSys 2025.
[Jul. 2024] Our paper is accepted to ACM MobiCom 2024.
[Dec. 2023] Our paper is accepted to IEEE INFOCOM 2024.
[May. 2023] Our paper is accepted to ACM MobiSys 2023.
[Dec. 2022] Our paper is accepted to IEEE INFOCOM 2023.

Publications

* Co-first author

Conference and Journal

[MobiCom 2025] Vega: Fully Immersive Mobile Volumetric Video Streaming with 3D Gaussian Splatting

Gunjoong Kim*, Seonghoon Park*, Jeho Lee, Chanyoung Jung, Hyungchol Jun, Hojung Cha
[MobiCom 2025] EOS: Energy-Optimized Super-Resolution on Mobile Devices for Live 360-Degree Videos

Seonghoon Park, Minchan Kim, Hyejin Park, Jeho Lee, Jiwon Kim, Hojung Cha
[EMSOFT 2025] Ember: Task Wakeup Sequence–Based Energy Optimization for Mobile Web Browsing

Seonghoon Park, Jiwon Kim, Jeho Lee, Hojung Cha
[EMSOFT 2025] SecureRide: Detecting Safety-threatening Behavior of E-Scooters Using Battery Information

Jiwon Kim, Geon Kim, Jeho Lee, Thiemo Voigt, Hojung Cha
[MobiSys 2025] ARIA: Optimizing Vision Foundation Model Inference on Heterogeneous Mobile Processors for Augmented Reality

Jeho Lee*, Chanyoung Jung*, Gunjoong Kim, Jiwon Kim, Seonghoon Park, Hojung Cha

🏆 Best Paper Award (2/233)
[MobiCom 2024] Panopticus: Omnidirectional 3D Object Detection on Resource-constrained Edge Devices

Jeho Lee, Chanyoung Jung, Jiwon Kim, Hojung Cha
[INFOCOM 2024] Vulture: Cross-Device Web Experience with Fine-Grained Graphical User Interface Distribution

Seonghoon Park, Jeho Lee, Yonghun Choi, Hojung Cha
[MobiSys 2023] OmniLive: Super-Resolution Enhanced 360° Video Live Streaming for Mobile Devices

Seonghoon Park*, Yeonwoo Cho*, Hyungchol Jun, Jeho Lee, Hojung Cha
[INFOCOM 2023] Crow API: Cross-device I/O Sharing in Web Applications

Seonghoon Park, Jeho Lee, Hojung Cha

Other Publications

[MobiSys 2025 Rising Stars Forum] Towards Accurate, Adaptive, and Real-time Machine Perception on Resource-constrained Platforms

Jeho Lee
[MobiSys 2025 Poster] Poster: Mixture of Class-aware Experts for Efficient AIoT Inference

Hyemin Jeong, Jeho Lee, Seunghyeok Jeon, Hojung Cha

Awards and Honors

Best Paper Award, ACM MobiSys 2025
SIGMOBILE Student Travel Grant, ACM MobiSys 2025
Rising Star, ACM MobiSys 2025

Projects

Panoptic3D: Multi-view, Multi-modal Dataset for Mobile 3D Perception

NuScenes-compatible mobile 3D perception dataset based on 360-degree camera, LiDAR, and IMU sensors

Project Page
Automatic On-browser GUI Segmentation

Browser extension-based system for automatic segmentation of web GUI through AI-driven page section detection and DOM-based post-processing

Project Page