COMPUTER VISION · SOFTWARE ENGINEERING
Luke Mikat
I work on computer vision and motion capture systems that connect cameras, sensors and the physical world.
About
I am a Media Informatics graduate (M.Sc.) working in computer vision, motion capture and software engineering. Most of my work involves systems that combine visual perception, sensor data and interaction with the physical world.
As a research assistant at IISY (Intelligent Interactive Systems Group) at the Berliner Hochschule für Technik, most of my work went into fitting parametric human body models to 3D scans — shape estimation with SMPL-X and MANO, later extended with per-vertex SMPL+D displacements. Alongside that came motion capture and human pose estimation from RGB cameras and IMU sensors, distributed camera setups and synchronization, SLAM-based tracking and VR environments for user studies. I also led the scanning and motion capture exercises in a Capture2Render master's course.
My Master's thesis covered personalized 3D hand reconstruction and video-to-texture mapping for use in human-computer interaction and robotic systems.
I like working between research and software engineering, turning computer vision and machine learning methods into software that holds up in practice.
TECHNICAL INTERESTS
COMPUTER VISION · SOFTWARE ENGINEERING
Experience
IISY (Intelligent Interactive Systems Group)
Berliner Hochschule für Technik
Student Research Assistant — Prof. Dr. Kristian Hildebrand
Berlin, Germany
June 2024 — July 2026
- Built scan2smpl, a pipeline for fitting parametric human body models (SMPL, MANO) to 3D scans, extended from MANO hands to full-body reconstruction with per-vertex SMPL+D surface detail.
- Developed computer vision and motion capture systems for research applications, integrating RGB cameras, IMU sensors and human pose estimation pipelines.
- Developed a distributed camera synchronization system for multimodal and wirelessly connected cameras, with hardware timing control in C on Teensy microcontrollers.
- Built research infrastructure including sensor data processing, SLAM-based smartphone trajectory reconstruction, Docker environments and VR study frameworks in C#.
- Led the scanning and motion capture exercises in the Capture2Render course, guiding students through full-body scanning, mocap recording and rigging their own avatars for use in interactive applications.
form4
Working Student — Full Stack Web Developer
Berlin, Germany
March 2022 — October 2022
- Developed and maintained client web applications in TYPO3, PHP, React and SQL, across frontend components, backend features and the database layer.
- Took features from ticket to production in GitLab workflows with Docker and CI/CD pipelines.
- Worked in a team that moved from ticket-based Waterfall processes to Scrum during my time there.
Projects
scan2smpl — Parametric Body Model Fitting
August 2026
Python • PyTorch • Open3D • SMPL-X • MANO
Fits SMPL-X and MANO to 3D scans from a handful of annotated landmarks, solving pose and shape against the full scan surface. Per-vertex SMPL+D detail brought surface error on a full body scan down to roughly 1.3 mm. Includes an annotation viewer and an automatic landmark quality check, so a fit is reproducible from its configuration.
Master's Thesis: 3D Hand Shape Estimation for Video-to-Texture Mapping
June 2026
OpenCV • Open3D • PyTorch
DFG-funded research project on personalized 3D hand reconstruction: estimates hand geometry from 3D hand scans and reconstructs hand motion from motion capture data. Textures are then mapped onto the reconstructed meshes by combining RGB and thermal imaging. Acquired a hand scan dataset in order to evaluate the method on robotic hand models.
Just Capture: Lightweight Framework for Synchronized Multi-View Data
April 2025
C • Java • Teensy • Ethernet
Research paper, second author. Synchronizes multimodal motion capture setups — industrial FLIR cameras, wireless Android clients and a Qualisys tracking system — with Teensy firmware for hardware triggering and network time sync.
VR Study Toolkit: Extension and Validation
January 2025
Unity • C# • Node.js • React • Redux
Master's project, team of three. Extended a VR research toolkit that logs and replays study data: migrated the study server frontend to React with Redux, added multi-client support and a live map of the study environment, then validated it by reproducing a published redirected-walking study in Unity.
IMU Tracking from a Rokoko Motion Capture Suit
July 2024
Python • Wireshark
Reverse-engineered the suit's network protocol by decoding the packets it broadcasts, recovering the raw IMU streams and feeding them into a custom tracking implementation that runs independently of the official software.
Cross Platform Mobile Application
April 2024
React Native • Expo • Firebase • SQLite
Offline-first multiplayer game for iOS and Android, now at over 8,000 users, with local SQLite persistence and Firebase cloud sync.
Image Similarity Search
February 2022
Python • Flask • OpenCV • Three.js
Bachelor's project, team of four. Interactive tool for exploring image datasets by visual similarity, with feature extraction and nearest-neighbour ranking behind a 3D browsing interface.
Education
M.Sc. Media Informatics
Berliner Hochschule für Technik (BHT)
April 2023 — July 2026
Focused on computer vision, machine learning, XR and DevOps-oriented software engineering, combining AI methods, sensor data and modern software development workflows.
Master's thesis: "3D Hand Shape Estimation for Video-to-Texture Mapping", written within a DFG-funded research project on personalized 3D reconstruction, computer vision pipelines and robotic applications.
B.Sc. Media Informatics
Berliner Hochschule für Technik (BHT)
October 2019 — March 2023
Covered software engineering, computer science fundamentals and interactive systems, with a focus on web engineering and application development.
The program covered web technologies and framework performance analysis, database systems, algorithms and data structures, computer graphics, UI/UX design and native application development.
My bachelor's thesis compared the performance of modern web engineering frameworks: their architectural approaches, runtime characteristics and suitability for different application scenarios.
Technical Skills
Programming Languages
The first three are what I work in day to day; the rest I have shipped with.
Computer Vision & ML
Perception and machine learning work, mostly from research projects.
Motion Capture, Sensors & XR
Connecting software to sensors, hardware and VR environments.
Web & Mobile
Frameworks I have used to build web and mobile applications.
Data & Backend
Storage and backend work behind the applications above.
Tools & Platforms
Development workflows and engineering infrastructure.
Contact
Interested in computer vision, software engineering, XR or the software around them? Contact me!
Happy to talk about research, new projects, or roles where software engineering meets computer vision and real-world applications.