AISEE Software Expert Search Tool for Videos and Photographs
AI-powered analytical accelerator for processing large volumes of video recordings and photographs in police operations and forensic practice.
Project no. VJ02010029
Duration 01/2022 – 12/2025
Funder Ministry of Interior CR — IMPAKT 1

About the Project
AISEE (Softwarový expertní vyhledávač pro videa a fotografie) is a software platform developed within a joint research project funded by the Ministry of Interior of the Czech Republic under the IMPAKT 1 security research programme. The project bridges cutting-edge AI research and the real operational needs of the Czech Police (PČR) and forensic experts.
The platform acts as an analytical accelerator — enabling investigators to efficiently search through massive archives of surveillance footage and photographs using natural language, image examples, or pre-defined object classes, without requiring specialist AI knowledge.
Key Goals
- Transfer of state-of-the-art AI research into real-world police operations and forensic practice
- Unique fine-tuning of new object classes within the PČR environment with minimal input and without requiring expert AI knowledge
- Software platform AISEE as an analytical accelerator for processing large volumes of video recordings and photographs
- Technical solution accompanied by analysis of legal, ethical, and human-rights aspects
- Improved efficiency through defined classes, text queries, image-pattern search, and subsequent logical combinations
- "Human-in-the-loop" principle — final verification and decision always rests with the human expert
AISEE Platform
- System properties include local deployment, asynchronous processing, a user-friendly UX/UI, robustness, and on-site fine-tuning.
- Query interface: The platform provides an intuitive interface for formulating queries via predefined object classes, free-text descriptions, or reference images. Results are shown as a grid with quick jump-to-time functionality in video.
- Object-of-interest detection: The detection engine resolves complex spatio-temporal relationships between objects and leverages rich attribute sets (colour, size, behaviour) to narrow down results.
- On-site AI training: New object categories can be trained directly within the PČR environment using weakly supervised learning, minimising the amount of labelled data required and involving end-users throughout the learning process.
- Use cases include situational awareness, effective intervention, crisis threat analysis, and relevant data extraction.
- Result filters include Region of Interest (ROI), sequence, time interval, motion, confidence threshold, and favourites.
AISEE — GUI/UX

AISEE interface: video source panel (left), detection timeline, and results grid with thumbnails sorted by confidence score.
Legal & Ethical Framework
The project addresses legal, ethical, and human-rights aspects as an integral part of its technical development, ensuring compliance with current EU regulation.
AISEE and the EU AI Act
- AISEE is assessed as a safe and compliant tool for forensic analysis
- Design emphasises human operator accountability at every decision point
- Fully respects the normative framework of the EU AI Act (AIA)
- Analyses performed by the Research Law Centre (RLC) of the Institute of State and Law, Czech Academy of Sciences
Research Team
ÚTIA AV ČR - Department of Image Processing
- doc. RNDr. Barbara Zitová, Ph.D. (PI)
- prof. Ing. Filip Šroubek, Ph.D. DSc.
- Ing. Adam Novozámský, Ph.D.
- Ing. Michal Bartoš, Ph.D.
- Ing. Babak Mahdian, Ph.D.
- Ing. Jan Kamenický, Ph.D.
- Ing. Vít Hanousek
- Bc. Michal. Průšek
- Bc. Petr Vojtášek
VUT Brno - Faculty of Information Technology
- doc. RNDr. Pavel Smrž, Ph.D.
- Ing. Markéta Juránková, Ph.D.
- prof. Dr. Ing. Pavel Zemčík
- Ing. Michal Hradiš, Ph.D.
- Ing. Roman Juránek, Ph.D.
- Ing. Marek Šolony, Ph.D.
ČVUT Praha - Centre for Machine Perception
- prof. Mgr. Ondřej Chum, Ph.D.
- Ing. Michal Neoral, Ph.D.
ÚSP AV ČR - Research Law Centre (RLC)
- Mgr. Alžběta Solarczyk Krausová, Ph.D., LL.M.
Contact
doc. RNDr. Barbara Zitová, Ph.D.
Department of Image Processing
ÚTIA AV ČR
Pod Vodárenskou věží 4
182 00 Praha 8