Work Experience

Research Developer

  • Czech Technical University in Prague, Faculty of Electrical Engineering
    February 2025 -- present
    Applied computer vision research and development across several grant- and industry-funded projects, centred on detection, classification and fine-grained recognition in images and video.
    More details in Projects below and in Publications

Research Scientist

  • Czech Technical University in Prague, Faculty of Electrical Engineering
    July 2017 -- January 2025
    Computer vision research for Toyota Motor Europe on dense motion estimation in monocular video: optical flow with explicit occlusion reasoning, segmentation of independently moving objects seen from a moving camera, and long-term tracking of every pixel. This line of work became my dissertation and four patent filings.

University Teacher

  • Czech Technical University in Prague, Faculty of Electrical Engineering
    March 2018 -- present
    Lab teaching for Pattern Recognition and Machine Learning (RPZ), Computer Vision Methods (MPV), and Digital Photography Processing (DIF). I also developed the automatic evaluation tool for student homework used in the RPZ course.
    More details in Teaching

Researcher and Developer

  • Czech Technical University in Prague, Faculty of Electrical Engineering
    2012 -- February 2017
    Part-time research and development alongside my bachelor's and master's studies, on the CloPeMa, SyRoTek, Electrolux and Toyota projects listed below.

Projects

  • Project: FunDive
    2026 -- present
    Open-set and closed-set recognition of fungal species. Classification is hard here because the visual differences between species are fine-grained, natural variability is high, the data is heavily imbalanced, and unknown species appear at inference time.

  • Project: Local CLS
    2026 -- present
    A multi-stage knowledge-distillation method for fine-grained image retrieval, including a learned-threshold attention pooling mechanism that produces compact descriptors from frozen vision-language models.

  • Project: CEDMO
    2026
    A user-facing tool for automated analysis of billboard video recordings, associating detected people and textual content across many captured instances of the same billboard to support large-scale media monitoring.

  • Project: AISee
    2025 -- 2026
    An AI surveillance system for large-scale video object detection and classification, with fast incremental learning of classes defined by the user.

  • Project: Median
    2025 -- 2026
    A system for automatic recognition of template videos in cam-glasses recordings.

  • Project: Toyota Lab - Computer vision research for Toyota Motor Europe
    2015 -- 2025
    The Toyota Research Lab at the Faculty of Electrical Engineering, CTU Prague. Dense motion estimation in monocular video: optical flow, occlusion estimation, moving object discovery and segmentation, and long-term dense tracking.

  • Project: Electrolux
    2015
    Collaboration on an identification system for objects inserted into home appliances.

  • Project: SyRoTek -- A Robotic System for Education
    Systém pro Robotickou Tele-výuku (CZE)
    2014 -- 2015
    SyRoTek is a robotic e-learning platform designed to provide remote access to real mobile robots for educational purposes. I was involved in developing software modules that allowed students to control robots within a simulated arena from web application.

  • Project: CloPeMa -- Clothes Perception and Manipulation
    2012 -- 2014
    The CloPeMa project focused on advancing the autonomous perception and manipulation of various fabrics and garments. I contributed to developing visual perception algorithms that enabled robots to recognize and handle different types of clothing, facilitating tasks like sorting and folding. This work became my bachelor thesis.
    Tools: C++, Python, ROS

Internship

  • Toyota Motor Europe (TME), Brussels, Belgium
    February 2017 -- June 2017
    Developed and implementation of monocular depth estimation.
    Tools: C++, CUDA, Python, CAFFE2, TensorFlow